Section 5: Gastroenterology

INTRODUCTION

Gastroesophageal reflux disease (GERD) and other esophageal disorders account for a significant number of physician office visits, hospital admissions, and endoscopic procedures each year. Annual management costs for GERD alone have been estimated to exceed $9 billion and continue to escalate. Gastroesophageal reflux disease results from the backflowing of gastroduodenal contents into the esophagus, leading to a variety of symptoms including heartburn, chest discomfort, regurgitation, and dysphagia. ­Compromise in the physiological barrier of reflux at the gastroesophageal junction is believed to be the primary mechanism of reflux. Esophagitis is defined as the presence of signs of inflammation of the esophageal mucosal surfaces, both macroscopically and microscopically. Gastroesophageal reflux disease is the most common cause of esophagitis, followed closely by medication-induced and infection-induced inflammation. Symptoms of esophagitis may range from mild chest discomfort to significant dysphagia and odynophagia.

EPIDEMIOLOGY

Epidemiological studies have shown that approximately 25% of the Western population report having symptoms of heartburn at least once a month, with up to 5% noting daily symptoms. While the prevalence of heartburn symptoms appears to be similar in both genders, female patients less likely possess objective findings of reflux on physiological studies. Moreover, numerous clinical studies have shown a lower response rate to antireflux treatment for female patients. No definite relationship between age and GERD has been concluded. At least one study has suggested less reflux symptoms in older individuals. However, elderly also seem to possess more severe esophagitis, suggesting a higher prevalence of asymptomatic reflux. An increase in body mass index both causes reflux and exacerbates existing symptoms.

PRESENTATION

Gastroesophageal reflux disease most typically presents as heartburn, regurgitation, and dysphagia. Patients often describe heartburn as a retrosternal, “burning” discomfort in the lower chest or epigastric region that often travels up toward the neck region. Symptoms often worsen after meals or when patients are in a supine position. As a result, patients may complain of symptoms or exacerbation of symptoms at night that may wake them up from sleep. Physiological studies suggest that gastric distention caused by meals and sleep can each induce transient relaxation of the lower esophageal sphincter (LES), leading to episodes of reflux. Dysphagia is a common symptom of GERD, and it is often a result of refluxate-induced esophageal mucosal inflammation or dysmotility. However, it is also one of the “alarming” signs of GERD that indicate the need for further evaluation to exclude possible underlying malignancy. Other “alarming” signs of GERD include odynophagia, weight loss, signs of gastrointestinal (GI) bleeding, advanced age, and a family history of upper GI tract malignancy. When one or more of the “alarming” signs are present, further evaluation with endoscopy or other radiographic imaging should be considered.

In addition to the typical esophageal presentation, GERD may also cause a variety of atypical, extraesophageal manifestations. Coexisting pulmonary disease with GERD occurs commonly, with the diagnosis established in 40% to 80% of asthmatics. Although airway irritation resulting from microaspiration of refluxed contents has been implicated as a possible trigger for asthma, certain medications used for asthma, such as bronchodilators, may also cause transient LES relaxation. Consistent treatment benefit for asthma has not been demonstrated using acid-suppression medications or antireflux surgery. Nevertheless, GERD likely plays a role in the pathogenesis of pulmonary disorders such as asthma, chronic bronchitis, bronchiectasis, aspiration pneumonia, and pulmonary fibrosis, as well as upper airway manifestations such as pharyngitis, laryngitis, or postnasal drip. Some studies have shown worsened clinical course of these pulmonary and airway disorders in patients with GERD. Oral problems such as dental erosions, halitosis, and aphthous ulcers have also been linked to GERD.

Patients with GERD or esophagitis may also present with noncardiac chest pain, often described by patients as nonspecific retrosternal discomfort. Such chest discomfort may be a result of direct irritation by the refluxate, mucosal inflammation, or esophageal spasm or dysmotility. Gastroesophageal reflux disease may mimic other causes of chest pain such as acute coronary syndrome or pulmonary disorders. Noncardiac chest pain resulting from esophageal disorders is a common presentation leading to emergency room visits and hospital admission for further workup.

PATHOPHYSIOLOGY

GERD

The lower esophageal sphincter, a tonic smooth muscle located in the lower esophagus around the gastroesophageal junction, serves as the major barrier to reflux of gastric content into the esophagus. Disruption in the tonic contractile pressure of the LES may result in the backflowing of gastric content into the esophagus. Such a disruption may present in the form of a general decrease in the tonic pressure of the LES (hypotensive LES) or inappropriate relaxation of the LES not during swallowing (pathologic transient LES relaxation). Many factors have been associated with disruption of the tonic contraction of the LES. For example, a hiatal hernia leads to anatomic disruption of the gastroesophageal junction and is associated with an increased risk of transient LES relaxation. Certain medications such as sedatives, food items such as caffeine and chocolate, and lifestyle behavior including alcohol and tobacco use can also lead to increased LES relaxation, resulting in GERD symptoms. In addition to LES relaxation, the barrier between the stomach and esophagus may also be pathologically overcome if the intragastric pressure exceeds the tonic contractile pressure of the LES. Most commonly, this situation results from obesity, delayed gastric emptying, or certain postoperative states, such as gastric bands for weight loss.

While GERD symptoms are often thought to be a result of the presence of acidic gastric content in the esophagus, leading to burning sensation and inflammation, the backflow of nonacidic contents such as bile may also generate similar symptoms. Since certain diagnostic modalities including pH study and the use of acid-suppression medications for treatment of GERD targets the presence of acid in the esophagus, nonacid reflux should be considered in patients who do not respond to common medical therapy for GERD symptoms.

ESOPHAGITIS

Esophagitis presents with inflammatory changes in the esophageal mucosa, with or without signs of erosion. The inflammatory reactions in the esophagus can be elicited by direct caustic injury, most commonly by acid, trauma, infection, or autoimmune processes. In the case of reflux-induced esophagitis, the diffusion of hydrogen ions into the esophageal mucosa leads to cellular acidification and necrosis. As a result, erosion and inflammatory cells infiltration ensue. Caustic injuries may also be caused by direct erosion or chemical reactions between the substance and the esophageal mucosa.

DIFFERENTIAL DIAGNOSIS

GERD

A variety of risk factors have been linked to GERD. Certain medications may cause a decrease in LES pressure, including β-adrenergic agonists, α-adrenergic antagonists, anticholinergics, benzodiazepines, calcium channel blockers, narcotics, progesterone, estrogens, and tricyclic antidepressants. Lifestyle behaviors such as smoking, alcohol ingestions, and food ingestion prior to incumbency have been linked to GERD symptoms. Some food items and beverages can also cause heartburn symptoms by reducing LES pressure. Examples include fatty food, chocolate, citrus fruits, tomato-based products, peppermint, vinegar, carbonated drinks, and caffeinated beverages.

In addition to lifestyle behavior and ingestions, other known risk factors for GERD include obesity, gastroparesis, certain surgical procedures, and hiatal hernia. Obesity may lead to an increase in intra-abdominal pressure that can cause LES relaxation. ­Gastroparesis causes intragastric distention, which may in turn activate stretch receptors of the stomach, resulting in decrease in LES pressure.

ESOPHAGITIS

Inflammation of the esophagus is most commonly related to GERD. Other risk factors for the development of esophagitis include medication, infection, radiation, or less common causes such as autoimmune disorders. Medication-induced esophagitis (“pill esophagitis”) results from direct, caustic injury to the esophageal mucosa. ­Common examples include alendronate, aspirin, nonsteroidal anti-inflammatory drugs, iron salts, potassium chloride tablets, quinidine, and tetracycline. However, any medication or foreign object can lead to direct injury and erosion into the esophageal mucosa if its transit through the esophagus is prolonged.

Infectious esophagitis usually occurs in immunocompromised hosts such as patients with acquired immune deficiency syndrome, post-transplantation, or undergoing chemotherapy. The most common etiologies for infectious esophagitis are fungal and viral infections. Esophageal candidiasis is one of the most common opportunistic infections in immunocompromised patients (­Figure 155-1). Often, it represents the initial presentation leading to the diagnosis of HIV infection, and candida esophagitis (unlike simple thrush alone) is an acquired immune deficiency syndrome defining illness. Patients with esophageal candidiasis usually complain of dysphagia and occasionally odynophagia. Esophageal candidiasis typically appears as whitish plaques on the esophageal mucosa that are easily scrapped off and oropharyngeal disease (thrush) often coexist. Therefore, in immunosuppressed patients who present with dysphagia, an oral examination may provide clues to possible esophageal candidiasis. Empiric treatment with antifungals may be considered in these high-risk patients. The most common agents causing viral esophagitis are human herpes simplex virus (HSV) and cytomegalovirus (CMV). Similar to fungal candidiasis, HSV and CMV esophagitis often present with dysphagia and odynophagia, although odynophagia presents more commonly in these cases. Concomitant oropharyngeal disease often occurs. Diagnosis requires endoscopic visualization of the typical HSV or CMV ulcers.

CANDIDA ESOPHAGITIS

Cancer patients with prior chest radiation therapy may also develop radiation-induced esophagitis. Symptom-onset can be immediate (within 3 weeks) or delayed (usually 3 or more months), with significant variation in the severity often unrelated to the dosage of radiation received. Symptoms range from mild dysphagia to severe odynophagia or stricturing requiring enteral tube feeding. As these patients are often also immunocompromised from chemotherapy, infectious esophagitis may coexist and must first be ruled out or treated.

Patients with a history of hematologic malignancy who have undergone bone marrow transplantation may develop ­graft-versus-host disease, which can involve any part of the gastrointestinal tract. Graft-versus-host disease of the esophagus often leads to significant inflammation and ulceration, resulting in symptoms of dysphagia, odynophagia, or bleeding. Therefore, this diagnosis must be considered in bone marrow transplant patients presenting with upper gastrointestinal bleeding or new-onset dysphagia symptoms, and mucosal biopsies should be obtained to establish the diagnosis. Similar to other cancer patients, infectious esophagitis is also very prevalent in this population and may coexist with graft-versus-host disease due to chronic immunosuppression.

Autoimmune disorders remain an uncommon cause of esophagitis. Inflammatory diseases such as Crohn’s disease or Behçet’s disease may involve the esophagus, leading to inflammation and often ulcerations. Eosinophilic esophagitis, another immune-­mediated esophageal disorder, appears to be increasing in incidence in recent years. It is characterized by the infiltration of large numbers of eosinophils (>15 per high-powered field) in the esophageal mucosa. Etiology of eosinophilic esophagitis remains unclear; however, it is often linked to other allergic disorders such as asthma or dermatitis. Patients with eosinophilic esophagitis are often young and they often first present with an episode of food impaction. While eosinophilic esophagitis can be associated with characteristic endoscopic appearances, its diagnosis requires histopathologic confirmation of eosinophilic infiltration. In recent years, a subset of patients with increased eosinophilic infiltration in the esophageal mucosa has been found to respond, both clinically and histologically, to proton pump inhibitor (PPI) therapy. This condition is now separately classified as PPI-responsive esophageal eosinophilia (PPI-REE). The etiology of PPI-REE is unclear, as most patients do not report GERD symptoms. Patients with PPI-REE are clinically, endoscopically, and histologically indistinguishable from patients with eosinophilic esophagitis, with response to PPI therapy being the only differentiating factor. Therefore, the most recent management guidelines from the American College of Gastroenterology suggest that patients found to have esophageal eosinophilia should first undergo a PPI trial of at least 8 weeks in duration. Repeat endoscopy with biopsies should then be performed to assess response to the PPI and classify patients into PPI-REE versus eosinophilic esophagitis.

DIAGNOSIS

ACID SUPPRESSION AS A DIAGNOSTIC TOOL

Gastroesophageal reflux disease diagnosis is most frequently achieved through history of classic symptoms combined with response to empiric antacid medical therapy, most commonly a PPI. A meta-analysis of studies on the diagnostic value of PPI response in GERD revealed a pooled sensitivity of 78% and specificity of 54% (likelihood ratio 1:70). In most cases, no further diagnostic testing is needed in patients with typical GERD symptoms who have no alarming signs that resolve with a therapeutic trial. However, patients who fail to respond to a trial of PPI or report alarm signs should undergo further diagnostic workup, including endoscopy, radiographic imaging, or physiologic testing such as manometry, impedance, and pH studies.

UPPER ENDOSCOPY

Upper endoscopy or esophagogastroduodenoscopy serves as an appropriate initial evaluation of GERD as it allows direct visualization of the esophageal mucosa to exclude the presence of other diseases such as tumors and evaluate for signs and severity of GERD-induced esophagitis. The Los Angeles classification, the most commonly used system in the endoscopic grading of severity of reflux esophagitis, measures the length and circumferential involvement of mucosal breaks and assigns the findings into four different grades (A-D). In addition to evaluation of esophagitis, endoscopy also allows the diagnosis of complications from GERD, including strictures or the presence of Barrett’s esophagus. Endoscopic dilation may be performed at the time of identification of strictures. Overall, endoscopy has been shown to exhibit high specificity (>90%) for the diagnosis of GERD but only a moderate sensitivity (~50%). The lack of endoscopic signs of GERD does not exclude its diagnosis. Previous studies have shown that many symptomatic patients with negative initial studies develop mucosal lesions during follow-up endoscopic examinations. Therefore, symptomatic patients with negative endoscopies should undergo further diagnostic workup.

RADIOGRAPHIC STUDIES

Radiographic studies such as barium swallow or upper GI series can also be used to detect esophagitis, hiatal hernias, esophageal strictures, or tumors. The primary goal of these studies is to exclude other significant diseases in patients with low clinical suspicion for them. Although they have the advantage of being noninvasive, inexpensive, and widely available, they are less sensitive and specific and should only be used over upper endoscopy if no endoscopist is available or if the patient cannot tolerate an endoscopic procedure.

MOTILITY STUDIES

Several kinds of motility studies have been used for the workup of reflux disease. Ambulatory esophageal pH monitoring records the distal esophageal pH continuously using a transnasal probe or a pill-sized capsule (Bravo capsule) placed 5 cm above the LES. A battery-powered device carried by the patient records data from the probe, as well as the time of meals and symptom-onset. This allows correlation between patient-reported symptoms and objective signs of acid exposure in the distal esophagus, described as symptom association probability (SAP) or symptom index (SI). SAP > 95% or SI > 50% indicates significant association between symptoms and acid exposure. A pH < 4 measured by the probe is defined as acid exposure. In normal individuals, the total amount of time with esophageal pH < 4 (total acid exposure time) over a day should be less than 4%. Although esophageal pH monitoring provides the most sensitive test for identifying presence of acid in the esophagus, its use should be reserved for selected patients: those with symptoms refractory to therapy, atypical symptoms unresponsive to acid suppression, or history of antireflux surgery, or those undergoing preoperative evaluation for antireflux surgery or lung transplantation.

Esophageal manometry measures the strength, duration, and sequential nature of esophageal smooth muscle contraction. An array of pressure sensors along a small catheter passed through the nose into the esophagus and stomach measures the esophageal body contraction, as well as the resting pressure and relaxation of the LES. With each swallow, different pattern of muscle contraction along the probe allows identification of the presence of any motility abnormalities, LES hypotension, or transient relaxation. This test is particularly important as a part of the preoperative evaluation for antireflux surgery. In addition to obtaining valuable information regarding LES pressure, manometry also identifies patients with underlying esophageal disorders such as achalasia or diffuse esophageal spasm that would be a contraindication to fundoplication.

Impedance monitoring is a technique in measuring bolus transit through the esophagus. Similar to manometry, it contains a small, flexible catheter placed through the nose into the esophagus. In GERD patients, impedance can identify fluid bolus refluxing into the esophagus, regardless of its acidity, and measure the extent of reflux. The combined impedance-pH monitoring device can further classify the refluxed fluid bolus as acidic, weakly acidic, or nonacidic. Newer versions of these catheters may also have additional sensors located more proximally, up to the oropharynx, to help identify proximal or pharyngeal reflux.

TRIAGE AND HOSPITAL ADMISSION

Although GERD and esophagitis can cause significant morbidity for patients, they are still generally managed in the outpatient setting. In rare instances, patients who cannot tolerate oral diet due to significant dysphagia or odynophagia caused by severe esophagitis may require hospitalization for nutritional support. These patients may demonstrate clinical signs of significant dehydration, including hypotension and tachycardia. In the setting of prolonged poor nutritional status, clinical signs related to deficiency of other nutritional elements such as folate and certain vitamins may also be present. In these cases, intravenous hydration need to be initiated, and other forms of nutritional support such as tube feeding or total parenteral nutrition may need to be considered.

Hospital admission in patients presenting with GERD or esophagitis-related symptoms may be required to rule out other significant illnesses. Certain GERD and esophagitis symptoms may mimic more significant cardiovascular diseases such as acute coronary syndrome. Patients presenting with chest discomfort should undergo careful cardiac evaluation, and the diagnosis of GERD and esophagitis should not be assumed until cardiac causes can be ruled out.

MANAGEMENT

SYMPTOMATIC MANAGEMENT

Treatment of patients with GERD or esophagitis should control the symptoms as well as heal esophageal inflammation. Management of GERD and reflux esophagitis focuses on eliminating acid exposure to the esophagus using both pharmacological (antacid and antisecretory agents) and surgical means. Over-the-counter antacids neutralize refluxed gastric acid, are inexpensive and convenient, and may be used as short-term treatment for patients with occasional symptoms.

H₂ blockers inhibit gastric acid secretion by competitively binding the H₂ receptors on gastric parietal cells. They prevent heartburn symptoms when taken before meals and relieve postprandial heartburn if taken after meals. In patients with frequent GERD symptoms, twice daily H₂ blockers may be required for maximal effects. H₂ blockers offer a 10% to 24% therapeutic gain over placebo in healing mild-to-moderate esophagitis, but are not effective in healing moderate to severe esophagitis. No differences in efficacy among the different H₂ blockers have been found. If a patient remains symptomatic after 6 weeks of standard dose, twice daily H₂ blocker therapy, increased dose or prolonged therapy do not appear to provide additional relief.

Proton pump inhibitors directly block the hydrogen-potassium ATPase on the apical surface of the parietal cells and act on a more downstream pathway than H₂ blockers to decrease acid secretion. Compared to H₂ blockers, PPIs have greater effectiveness in treating esophagitis after 4 to 8 weeks of therapy. All first-generation PPIs (all except esomeprazole) have similar effects on healing of esophagitis and gastric acidity. Esomeprazole has been shown to have a statistically significant advantage in average gastric pH and percentage of time with nonacidic pH when compared to other PPIs after 5 days of therapy. However, no objective evidence has demonstrated any significant benefit of esomeprazole over other PPIs in symptoms-control or esophagitis healing. The timing of PPI use is important as its efficacy depends on the presence of proton pumps on the parietal cells. Therefore, PPIs should be taken during fasting state approximately 30 minutes before meals to allow for maximal effects.

While short-term side effects associated with PPI use are rare and relatively mild, increasing evidence have emerged over the past few years suggesting that prolonged use of PPI may have long-term, systemic consequences. Malabsorption, including vitamin B12 and vitamin D, osteoporosis, Clostridium difficile colitis, community-acquired pneumonia, and symptomatic small intestinal bacterial overgrowth have all been linked to chronic PPI use. Concerns regarding absorption and treatment effects of certain medications have also been raised in patients with concurrent PPI use. In particular, the concomitant use of PPI and clopidogrel has been under intense scrutiny given the large number of cardiac patients taking clopidogrel who also require PPI therapy. Several in vitro studies have demonstrated evidence of a decrease in efficacy of clopidogrel with PPI using serum markers for platelet response, leading the Food and Drug Administration (FDA) to issue a warning advising the avoidance of certain PPI and clopidogrel coadministration. A large number of nonrandomized, observational clinical studies have been published, with the results ranging from no clinical effect to a mild-to-moderate increase in clinical events such as recurrent MI or hospitalization. The only large, randomized trial studying the effect of PPI (omeprazole) versus placebo on clinical outcome in 3761 patients taking clopidogrel (COGENT) demonstrated no difference in cardiovascular events (HR 1.02, 95% CI: 0.70-1.51), but a significant decrease in GI events (HR 0.55, 95% CI: 0.36-0.85). Furthermore, a meta-analysis published in April, 2010 showed no significant association between concomitant PPI-clopidogrel use and overall mortality (RR 1.09, 95% CI: 0.94-1.26). A more recent systematic review and meta-analysis published in January, 2015 demonstrated higher ischemic event rates with coadministration of clopidogrel with PPI (as a class) among observational studies, but no increase in ischemic events among four randomized controlled trials studying omeprazole-clopidogrel coadministration. Given the short half-life of both PPI and clopidogrel, some experts have suggested that separating their intake by 12 to 20 hours (eg, one taken before breakfast and the other before bedtime) should, in theory, minimize any potential clinical interaction. Regardless, the use of PPI should be judicious in light of evidence of potential long-term side effects, and dose-reduction or discontinuation should be considered after symptom-control or mucosal healing has been achieved.

Other pharmacological agents studied for the management of GERD, especially in those with nonacidic or weakly acidic reflux, include baclofen, which has been found to be effective in inhibiting transient lower esophageal sphincter relaxation, thereby decreasing acid reflux. However, the use of baclofen has been limited by its short half-life and poor patient tolerability.

While most patients achieve adequate treatment response with pharmacotherapy, some may experience symptoms refractory to acid suppression or wish to discontinue medication use. In these cases, barrier enhancing antireflux therapy either surgically or endoscopically can be attempted. Antireflux surgeries have been safely performed since the 1950s, and the advance of laparoscopic approach further decreases the complication rate and postoperative hospital stay. ­Nissen fundoplication, either open or laparoscopic, involves enhancing the pressure at the LES region by wrapping and attaching part of the gastric fundus around the distal esophagus. Studies have shown a symptom response of 80% to 90% postoperatively. However, long-term studies demonstrated that the majority of patients experience recurrent symptoms and require acid-suppression medications again within 10 years. Complication rate of fundoplication ranges from 5% to 20%, most commonly due to dysphagia, chest pain, gas-bloat syndrome, flatulence, and vagal nerve injuries. Dysphagia remains the most commonly encountered complication from fundoplication. It may occur early in 0% to 20% of patients, or late in up to 6% of patients at 2 years. Preoperative physiologic studies including manometry and pH testing can help identify patients with higher likelihood of positive or negative outcomes. Increased LES relaxation and positive pH testing results are positive predictors of fundoplication outcome, while patients with normal pH testing or normal LES pressure are less likely to respond to fundoplication. In addition, patients with advanced spastic esophageal disorders or signs of achalasia on manometry are contraindicated for surgery given the high risk of postoperative dysphagia.

A new surgical antireflux procedure has been approved by the FDA involving the placement of a bracelet with magnetic beads at the gastroesophageal junction. Magnetic force approximates the beads and tightens the bracelet, thereby enhancing the barrier at the gastroesophageal junction against refluxing of gastric contents. These magnetic beads are separated by the propulsive force from esophageal peristalsis during swallows, allowing the opening of the bracelet and passage of ingested food bolus through the gastroesophageal junction. Clinical studies have demonstrated this procedure to be effective and a safe alternative to traditional fundoplication in selected patients, although further studies are needed to assess the long-term efficacy and safety of this procedure.

Obesity has been increasingly recognized as an etiologic factor in GERD. Weight loss, through both surgical and nonsurgical strategies, has been shown to improve GERD and related symptoms. Therefore, in obese patients, weight loss surgery with Roux-en-Y gastric bypass may be preferred over traditional antireflux surgery in management of GERD symptoms. The selection of the type of bariatric surgery in GERD patients should be careful, as certain strategies, particularly gastric banding, have been associated with increased reflux.

NUTRITIONAL SUPPORT

In rare cases of patients with significant esophagitis causing severe dysphagia or odynophagia, nutritional support may be needed due to their inability to tolerate adequate oral diet or hydration. Enteral feeding and hydration through a gastrostomy or jejeunostomy tube may be required. Intravenous hydration or parenteral nutrition may be needed in certain patients with severe malnutrition or other concurrent gastrointestinal disorder. Clinicians should pay particular attention to any signs of deficiency of specific elements such as vitamin B12, vitamin D, and calcium and the need for replacement.

LONGITUDINAL/OUTPATIENT THERAPEUTICS

Outpatient care of patients with GERD or esophagitis focuses on medical treatment of underlying disorders as well as lifestyle modification. Acid-suppression medications, such as PPI or H₂ receptor antagonists, are the mainstay of therapy as described above. Once symptoms resolve or esophagitis heal, dosages of acid-suppression medications may be slowly tapered as tolerated. Lifestyle modifications remain a cornerstone of GERD treatment. Avoiding foods and acid-containing beverages that exacerbate their symptoms should be advocated. Patients should be educated about not lying down after eating for at least 3 hours to avoid refluxing of ingested food. Similarly, eating smaller and frequent meals can help decrease GERD symptoms. For patients with significant nocturnal reflux, elevation of the head of the bed to 30^o to 45^o angle can help reduce nocturnal symptoms. Such elevation may be achieved by placing blocks underneath the front feet of the bed or triangular wedge underneath the mattress. In addition, avoidance of alcohol or tobacco use may reduce GERD symptoms or esophagitis. Weight loss in overweight or obese patients is also a significant part of GERD management.

INFECTIOUS ESOPHAGITIS

Treatment strategy for infectious etiologies of esophagitis can be broadly divided into acute eradication and prevention of recurrence. In patients with acute, symptomatic infectious esophagitis, antimicrobial therapy should be initiated and targeted toward the organism responsible. Determination of the duration of therapy, mode of administration (oral vs parenteral), and choice of antimicrobial agent should depend on a combination of factors including the patients’ symptoms, other medical comorbidities, and underlying immune function. Intravenous administration may be needed in those with significant odynophagia or patients who are acutely ill or severely immunosuppressed. Increased treatment duration and more potent agents should also be considered in these patients.

Patients with suspected or diagnosed esophageal candidiasis may be initially treated with a 14-day course of fluconazole (400 mg loading dose on first day followed by 200 mg daily on subsequent days), administered orally or intravenously. Fungal esophagitis unresponsive to fluconazole may require more potent antifungals such as amphotericin and should be managed with input from infectious diseases specialists. First-line therapy for HSV esophagitis in immunosuppressed patients involves a course of acyclovir (5 mg/kg IV every 8 hours for 7-14 days or 400 mg PO five times daily for 7-10 days), while CMV esophagitis requires the use of ganciclovir (5 mg/kg IV every 12 hours or 900 mg daily or twice daily) for 3 to 6 weeks. More potent antivirals such as foscarnet may be necessary in cases unresponsive to first-line agents and infectious diseases specialists should be involved in management of these cases. Herpes simplex virus or CMV esophagitis in immunocompetent hosts usually resolve without treatment after 1 to 2 weeks. Experience with orogenital HSV suggests that antiviral treatments may hasten symptom recovery. Therefore, some clinicians suggest treating immunocompetent patients with a shorter course of therapy.

COMPLICATIONS AND PROGNOSIS

Complications from chronic GERD can be divided by the sites of manifestations: esophageal versus extra-esophageal. Esophageal complications include ulcers, erosions, strictures, Barrett’s esophagus, and malignancy. Extra-esophageal complications mainly involve upper airway or pulmonary manifestations of aspirated gastric contents.

Esophageal ulcers may result from esophagitis caused by chronic GERD or impacted foreign bodies. They may account for up to 2% of all upper GI bleeding. However, the vast majority heal completely with acid suppressive medications. Chronic GERD and esophagitis may also lead to peptic esophageal strictures, which have been described in up to 10% of patients seeking medical attention for reflux symptoms. Patients with strictures usually present with dysphagia to solid food but can tolerate fluid without difficulties. Weight loss is uncommon and malignancy needs to be ruled out when present. Treatment for peptic strictures involves a combination of acid-suppression therapy and endoscopic dilation. Food and foreign body impaction is another common complication for patients with esophageal strictures or significant esophagitis. In addition to reflux esophagitis, another form commonly associated with food impaction is eosinophilic esophagitis. Treatment strategies for eosinophilic esophagitis include the use of topical steroids (swallowed fluticasone or budesonide slurry) or an elimination diet.

Chronic GERD may also predispose patients to the development of Barrett’s esophagus and esophageal adenocarcinoma. Barrett’s esophagus is the metaplastic change of the esophageal lining due to chronic reflux injury and esophagitis. It involves the replacement of the normal squamous epithelium of the distal esophagus with specialized intestinal metaplasia. Such metaplastic change may protect the esophagus against further injury by gastric acid, but it also increases the risk for esophageal adenocarcinoma. The prevalence of Barrett’s esophagus in GERD patients is approximately 5% to 10%. Caucasians, male, and middle-aged patients are more commonly affected. Other risk factors for Barrett’s esophagus include tobacco and alcohol use. The exact risk of esophageal adenocarcinoma in patients with Barrett’s esophagus remains debated, but is believed to be around 0.5% to 1%. The use of esophagogastroduodenoscopy to screen for Barrett’s inpatients with a history of GERD symptoms remains controversial, as there is no direct clinical data to support the utility of screening. The ­American ­Gastroenterological Association recommended against screening of the general population with GERD for Barrett’s in its most recent medical position statement published in 2014. However, it suggested screening for Barrett’s among GERD patients with multiple risk factors for esophageal adenocarcinoma (male, age >50, white race, chronic GERD, hiatal hernia, elevated body mass index, and intra-abdominal distribution of body fat). Surveillance for signs of dysplastic changes should be performed periodically if a diagnosis of Barrett’s has been confirmed on biopsy. Conservative treatment with continued surveillance remains the mainstay of therapy for Barrett’s without dysplasia. For those with high-grade dysplasia present on biopsies, treatment strategy includes endoscopic ablative therapy or surgical resection. Treatment strategy for low-grade dysplasia remains controversial but involves ablative therapy or conservative surveillance. Acid control should be continued as part of medical therapy for Barrett’s esophagus.

Extraesophageal complications of GERD may manifest in many forms, mainly resulting in pulmonary or upper airway abnormalities. GERD is believed to play a role in the pathogenesis of pulmonary disorders such as asthma, chronic bronchitis, bronchiectasis, aspiration pneumonia, and pulmonary fibrosis. Studies have also linked GERD to upper airway manifestations such as pharyngitis, laryngitis, or postnasal drip. Oral problems such as dental erosions, halitosis, and aphthous ulcers have also been associated with GERD. However, despite evidence of increased prevalence of these disorders in patients with GERD, treatment with acid-suppression medications has not consistently demonstrated clinical improvement in patients with these manifestations. Despite the lack of clear evidence, current clinical practice usually involves a workup for GERD in patients presenting with these disorders of unclear etiology or refractory to standard treatment.

The main goal of GERD and esophagitis treatment is to keep patients symptom-free with the least amount of acid-suppression medications that abates symptoms. Therefore, in patients who have responded to medical therapy or lifestyle changes, a trial of tapering and withdrawing medications should be attempted. In general, symptoms relief should be sustained for at least 2 to 3 weeks prior to the start of dosage tapering. Tapering should involve the sequential decrease in dosage or frequency for PPI, which may ultimately be switched to H₂ receptors antagonists. Abrupt withdrawal of PPI should be avoided as “rebound hypersecretion” may ensue when parietal cells secrete an elevated amount of acid after prolonged blockade. Adhering to lifestyle or dietary choices that decrease symptoms usually allows most motivated GERD patients to wean off medications. However, in patients with erosive esophagitis, relapse of symptoms and mucosal damage occurs in more than 80% of patients when therapy is stopped.

Patients with infectious esophagitis often respond well to first-line antimicrobial therapy. Those at highest risks of nonresponse are usually patients with profound immunosuppression. Clinical studies have demonstrated a decrease in Candida-, CMV-, or HSV-related diseases in immunosuppressed patients receiving respective prophylactic therapy. However, given the high acute treatment response rate, ease and low cost of therapy, and risk for development of drug resistance and sensitivity, long-term maintenance or secondary prophylaxis is not routinely recommended for infectious esophagitis, even in immunosuppressed patients. Some experts suggest consideration for maintenance therapy only in immunosuppressed patients with severe or frequent relapses of their infectious esophagitis.

DISCHARGE PLANNING

In the rare instances where patients were admitted for complications from their GERD or esophagitis, discharge planning mainly involves ensuring proper follow-up with a gastroenterology specialist, continuation of medical therapy, and patient education for secondary prevention. For patients with significant nutritional deficits requiring feeding tube placement, discharge planning should also include education on the proper use of the feeding tube, home nursing care or transfer to skilled nursing facilities, adequate equipment for continuation with tube-feeding after discharge, and instructions regarding caloric count. Follow-up for these patients should also include visits with a nutritionist for continued management of tube-feeding regimen. Educational materials regarding the prevention of recurrent GERD and esophagitis such as lifestyle modification strategies should be given, targeting their underlying etiologies.

QUALITY IMPROVEMENT

Secondary prevention

Secondary prevention plays a big role in the outpatient management of patients with GERD and esophagitis. Lifestyle modifications such as dietary and sleeping habits have been shown to be effective in decreasing symptoms in patients with GERD. In addition to providing educational materials, assistance from nutritionists and social workers may help patients achieve their lifestyle modification goals. Continued use of medications such as acid-suppression drugs may help prevent recurrent symptoms or development of complications. However, patients who adhere to lifestyle changes and achieve symptom relief should attempt a trial of medication weaning to allow maintenance on the lowest dose of medications possible. In patients with esophagitis, endoscopic surveillance to evaluate for mucosal healing aid in adjustment of medical therapy and in the evaluation of treatment efficacy, especially in other forms of esophagitis.

Transition of care

Patients with an initial diagnosis of GERD or esophagitis should continue care with a gastroenterology specialist for management of their diseases until symptoms resolve or condition stabilizes. Continued consultation with nutrition specialists may be needed for selected patients with ongoing GERD symptoms or difficulties adhering to lifestyle modifications. Certain underlying causes of esophagitis may require further care by other specialists, for example, patients found to have infectious esophagitis such as candidiasis or HSV may require further workup for underlying immunosuppression. Consultation with an allergy/immunology specialist should also be considered in patients diagnosed with eosinophilic esophagitis.

Outcomes to monitor

Symptoms resolution remains the main outcomes for monitoring in patients with GERD and esophagitis. Esophageal mucosal healing in patients with esophagitis, or endoscopic and pathologic evidence of resolution of underlying disease in patients with eosinophilic esophagitis, or Barrett’s esophagus serve as other outcomes for monitoring and surveillance.

Costs and resource utilization

As the most common gastrointestinal disease, GERD also represents the most costly digestive disorder. The annual management cost of GERD has been estimated to exceed $9 billion. Much of the costs may be related to overuse of acid-suppression medications or unnecessary endoscopic or other diagnostic procedures. In both the primary care and inpatient settings, there have been signs of PPI overuse in the treatment of nonspecific gastrointestinal complaints. In addition, inappropriate use of high-dose PPI and lack of tapering or withdrawal strategies result in many patients being kept on more acid suppressive medications than they require. As stated above, empiric PPI should be used only in patients with signs and symptoms suggestive of GERD or esophagitis. Atypical symptoms or complaints should be thoroughly evaluated before PPI therapy is started. PPI tapering and withdrawal should be attempted in all patients with controlled symptoms. Lifestyle modification should be better promoted as its success often complements the tapering of medications. Endoscopic and other diagnostic procedures should be reserved for patients with atypical or alarming symptoms, or those who failed empiric PPI therapy (Table 155-1). ­Adherence to this strategy allows the delivery of the most cost-effective care for one of the most common and costly disorders today.

SUGGESTED READINGS

INTRODUCTION

Upper gastrointestinal (GI) bleeding is responsible for over 300,000 hospitalizations per year in the United States. An additional 100,000 to 150,000 patients develop upper GI bleeding during hospitalizations. The annual cost of treating nonvariceal acute upper GI bleeding in the United States exceeds $7 billion.

Upper GI bleeding is defined as a bleeding source in the GI tract proximal to the ligament of Treitz. The presentation varies depending on the nature and severity of bleeding and includes hematemesis, melena, hematochezia (in rapid upper GI bleeding), and anemia with heme-positive stools. Bleeding can be associated with changes in vital signs, including tachycardia and hypotension including orthostatic hypotension. Given the range of presentations, pinpointing the nature and severity of GI bleeding may be a challenging task.

The natural history of nonvariceal upper GI bleeding is that 80% of patients will stop bleeding spontaneously and no further urgent intervention will be needed. In contrast, only 50% of patients with a variceal hemorrhage stop bleeding spontaneously. Following cessation of active variceal bleeding, there is a high risk of recurrent bleeding within 6 weeks.

The mortality rate for nonvariceal upper GI bleeding is 2% to 14%. This mortality rate has improved because of the development of new medications, endoscopy (both diagnostic and therapeutic), intensive care units (ICUs), and advances in surgical management. The mortality remains high since patients with GI bleeding now are older, have more comorbidities, and are taking more medications, including nonsteroidal anti-inflammatory drugs (NSAIDs), anticoagulants, and antiplatelet agents. For variceal bleeding, mortality is between 15% and 50% for each bleeding episode, and 70% to 80% in those with continuous bleeding. Variceal hemorrhage is responsible for one-third of all deaths due to cirrhosis.

DIFFERENTIAL DIAGNOSIS

NONVARICEAL BLEEDING

Peptic ulcer disease

Diagnosis: Peptic ulcer disease is a common condition stemming from an imbalance of protective and disruptive factors of the GI mucosa. Ulcers are most commonly found in the stomach and proximal duodenum (Figures 156-1 and 156-2). Peptic ulcer disease is the most common cause of upper GI bleeding and accounts for up to 50% of total cases and over 100,000 hospital admissions per year in the United States. The annual incidence of peptic ulcer disease in patients infected with Helicobacter pylori is about 1% per year, which is six to ten times higher than patients who are uninfected. Diagnosis is made by endoscopy performed when symptoms prompt endoscopic investigation. An initial step in management is to identify H. pylori infection and users of NSAIDs. These two risk factors account for the vast majority of ulcers in the upper GI tract.

Disruptive factors that can damage the mucosa of the GI tract include acid, pepsin, bile salts, ischemia, and H. pylori. Exogenous causes are predominately medications (NSAIDs, aspirin, and SSRIs). Stress-induced ulcers are a common cause of bleeding in patients hospitalized for other severe illnesses. Risk factors for stress-induced ulcers include respiratory failure (especially intubated patients), coagulopathy (international normalized ratio [INR] > 1.5 or platelets <50,000/μL), trauma, sepsis, renal failure, burns, ICU admission, and surgery.

Pathophysiology: The defensive forces of the esophagus include esophageal motility with clearance of refluxed materials, the lower esophageal sphincter to prevent reflux, and salivary secretions that contain bicarbonate. Gastric protective factors include the mucous layer as well as tissue mediators. H. pylori is a flagellated bacterium with several unique attributes that aid in its ability to break down gastric mucosa. These factors include its production of urease, which neutralizes the acidic gastric environment. NSAIDs inhibit cyclooxygenase, thus decreasing mucosal protective prostaglandin levels in the stomach. They also increase gastric vascular endothelial adhesion molecules, thus leading to neutrophil adherence and mucosal injury.

Clinical findings: Peptic ulcer disease presents in a variety of ways. Patients can experience epigastric pain, melena, hematochezia, dyspepsia, bloating, or can be asymptomatic with or without anemia. On examination, pain may be epigastric and reproducible, but many patients have right-sided pain or no pain at all.

Prognosis: Response to treatment and outcomes vary for peptic ulcer disease. Treatment is with proton-pump inhibitor (PPI) therapy until asymptomatic and/or mucosal healing is demonstrated via an esophagogastroduodenoscopy (EGD) and/or antibiotics where indicated and removal of offending agents (eg, NSAIDs) leads to mucosal healing (Table 156-1). Chronic, nonhealing gastric ulcers should prompt consideration of underlying illness or an alternative diagnosis, such as malignancy, and thus should be biopsied.

Mallory-Weiss tears

Diagnosis: Mallory-Weiss tears occur at the gastroesophageal junction, in the distal esophagus, or proximal stomach (Figure 156-3). Typically, these tears form after repeated and severe vomiting or retching. The tears may extend into the underlying blood vessels. Acute upper GI bleeding is the major clinical finding, which may be associated with epigastric pain, which may radiate to the back. Endoscopy is used to document the presence of a gastroesophageal tear and allows the possibility of therapeutic intervention.

Bleeding occurs when the tear involves the underlying esophageal venous or arterial plexus. Patients often have a history of nonbloody vomiting or retching before the onset of hematemesis. The tears are located at the esophagogastric junction, and can be within a hiatal hernia. Tears can extend downward into the cardia or, less commonly, upward into the esophagus.

Pathophysiology: Mallory-Weiss syndrome is characterized by longitudinal mucosal lacerations in the distal esophagus and proximal stomach, which are usually associated with forceful retching.

Prognosis: Although 40% to 70% of patients with bleeding Mallory-Weiss tears require blood transfusions, most tears heal spontaneously within 24 to 48 hours of presentation. Risk factors for rebleeding include portal hypertension and coagulopathy. Patients with active or ongoing bleeding can be treated with endoscopic hemostatic methods. For the rare patient with ongoing bleeding unresponsive to endoscopic control, intravenous (IV) infusions of octreotide, arterial embolization via angiography, and esophageal balloon tamponade have been used. In refractory bleeding, surgery with over-sewing of the bleeding vessel may be required.

Dieulafoy lesion

Essentials of diagnosis: A Dieulafoy lesion is a dilated, aberrant submucosal vessel that erodes through normal epithelium and is not associated with an ulcer (Figure 156-4). Dieulafoy lesions are typically located just below the gastroesophageal junction on the lesser curvature of the stomach but can be found in areas throughout the GI tract, including the esophagus, duodenum, and colon. When a Dieulafoy lesion bleeds, massive arterial bleeding can occur. In the absence of active bleeding, the lesion is difficult to see as it may appear as a small and subtle raised area, such as a nipple or visible vessel without an associated ulcer. In patients with massive bleeding and no obvious cause, the lesser curvature of the stomach within 6 cms of the gastroesophageal junction should be carefully inspected endoscopically for evidence of a Dieulafoy lesion.

The typical patient with a Dieulafoy lesion is a man, with multiple comorbidities, already hospitalized for other problems. Diagnosis is confirmed with finding a visible vessel or active arterial pumping, without an associated ulcer or mass. The diagnosis can be difficult, as bleeding may be massive such that the area may be covered with blood and the source may not be visualized. It is especially challenging to diagnose Dieulafoy lesions that are not bleeding as the absence of an associated ulcer makes detection of the vessel difficult.

Pathophysiology: The cause of a Dieulafoy lesion is unknown. The arteries are 1 to 3 mm in size, which is approximately 10 times the caliber of mucosal capillaries. The typical location is in the upper stomach along the high lesser curvature, within 6 cm of the gastroesophageal junction.

Clinical findings: A Dieulafoy lesion often presents with a massive acute upper GI bleed, although the bleeding may be self-limited in nature. The diagnosis is confirmed with finding a visible vessel or with active arterial pumping without an associated ulcer or mass.

Prognosis: Endoscopic treatment is the mainstay for bleeding control of a Dieulafoy lesion via cautery, clips or the combination of epinephrine injection and one of these two treatment modalities. The risk of recurrent bleeding following endoscopic treatment is relatively high due to the size of the underlying artery. In patients with recurrent bleeding, repeat endoscopic therapy is often the next step; however, IR angiography or surgical therapy may be necessary and consultation by a surgeon and interventional radiologist early on is recommended.

VASCULAR MALFORMATIONS

Angiodysplasia

Diagnosis: Angiodysplasias may occur throughout the GI tract, may be multiple in one gastrointestinal region, or coexist in several different gastrointestinal locations. Although angiodysplasias are a relatively common source of small bowel bleeding, they are a rare source of massive bleeding from the stomach. Bleeding from an upper GI vascular malformation is typically intermittent and low grade in nature, although it can be acute and overt.

Pathophysiology: The cause of most vascular malformations is unknown. Vascular malformations can be present at birth, although they do not appear to be hereditary. They occur equally in both sexes and in different races. Vascular malformations may be caused by a rupture or clotting of a blood vessel during fetal development, although not associated with other problems at birth.

Clinical findings: Bleeding from upper GI vascular malformations is typically intermittent and low grade in nature, although it can be acute and overt. The endoscopic appearance may be variable, but includes a superficial collection or tuft of blood vessels.

Prognosis: Endoscopic therapy with thermal coagulation is the treatment of choice with cautery applied first to the periphery and then to the center of the lesion. Argon plasma coagulation is most commonly used to treat vascular malformations and is an effective therapy. Certain medications have been utilized in an attempt to treat vascular malformations in the upper gastrointestinal tract. Combination estrogen/progesterone therapy has been widely used, as has octreotide. Angiography or surgery may be offered for failures of endoscopic and medical therapy, but is rarely necessary.

Gastric antral vascular ectasia

Diagnosis: Gastric antral vascular ectasia (GAVE), also known as “watermelon stomach,” is an idiopathic condition often associated with portal hypertension. Gastric antral vascular ectasia is a relatively unusual cause of upper GI bleeding. The bleeding that occurs is typically chronic and low grade, often associated with iron deficiency anemia and occult bleeding. It is unusual to have acute or massive upper GI bleeding from GAVE.

Pathophysiology: The etiology of watermelon stomach is not known; however, it has been theorized to be due to gastroduodenal prolapse and occurs in patients with cirrhosis, portal hypertension, and systemic sclerosis. Patients with concomitant portal hypertension have bleeding that is more difficult to control, requiring treatment of the underlying portal hypertension.

Clinical findings: During endoscopy, vascular malformations occur in rows of reddish stripes that radiate outward from the pylorus, in a pattern that resembles the stripes on a watermelon; hence, this entity has been referred to as “watermelon stomach.” These stripes represent rows of vascular malformations and the entity is easy to recognize in the classic form. However, GAVE may have a less organized appearance especially in patients with portal hypertension. Although the diagnosis is by endoscopy, it can be confirmed by biopsies obtained during endoscopy. GAVE histology shows areas of vascular ectasia associated with spindle cell proliferation and fibrohyalinosis.

Prognosis: Most patients will respond to endoscopic therapy most commonly with argon plasma coagulation. Several such endoscopic treatment sessions may be required. Radiofrequency ablation may be used to treat refractory cases. However, surgery with an antrectomy may be required for patients who have persistent bleeding despite endoscopic management.

Aortoenteric fistula

Diagnosis: Aortoenteric fistulas occur when there is a direct communication between the aorta and the gastrointestinal tract and can lead to massive bleeding. Patients may have a “herald” bleed with initial hematemesis or hematochezia, which may spontaneously remit. This might then be followed by torrential bleeding, including exsanguination, making it important to quickly diagnose this condition. Intermittent bleeding can occur if a blood clot overlies a fistulous connection. There may be abdominal or back pain (which occurs in about half of patients), or signs of fever and infection. On physical examination, a bruit may be heard and a pulsatile mass may be palpable at any point along the midabdomen representing the course of the abdominal aorta.

A high index of suspicion is needed to promptly diagnose a patient with an aortoenteric fistula for successful management. In a patient who has a known risk factor such as previous aortic grafting, this diagnosis should be considered early in the course. In fact, it should always be suspected in patients with massive bleeding from the second to third portion of the duodenum. The diagnosis can be confirmed by computed tomography with angiography or with traditional angiography. Diagnostic tests need to be done in an expeditious manner along with prompt vascular surgical evaluation as the management of an aortoenteric fistula is surgical. The mortality rate of an unrecognized fistula is 100%, whereas the mortality for a recognized fistula is 30%.

Hemobilia

Diagnosis: Hemobilia is a rare cause of acute upper GI bleeding that occurs from the hepatobiliary tract in less than 1% of patients who undergo a procedure to the biliary tree (endoscopic retrograde cholangiopancreatography [ERCP], surgery, or CT-guided procedures) and almost never spontaneously with the rare exception of invasive tumors. The classic triad of hemobilia is biliary colic, obstructive jaundice, and gastrointestinal bleeding. The diagnosis of hemobilia can be difficult by upper endoscopy, which may not clearly visualize the ampulla. A side-viewing duodenoscope is often utilized to visualize and access the ampulla directly or to perform ERCP.

Hemobilia should be considered in patients who have had a recent history of hepatic or biliary tract injury. This includes trauma to the area, percutaneous (or transjugular) liver biopsies, percutaneous transhepatic cholangiograms, ablations of liver tumors or biopsies. Hemobilia can also occur secondary to gallstones, cholecystitis, hepatobiliary tumors, hepatic abscesses, and aneurysms. Obstructive jaundice may be associated with biliary sepsis. Patients with these findings should be considered as potentially having hemobilia. Treatment is directed at the primary cause of bleeding, although ERCP may also be needed to remove clot from the bile duct causing obstruction. While treatment may occasionally be performed endoscopically, it usually needs to be done angiographically (such as by arterial embolization) or surgically.

Hemosuccus pancreaticus

Diagnosis: Hemosuccus pancreaticus is another rare cause of upper GI bleeding that occurs due to bleeding from the pancreatic duct. This can be secondary to chronic pancreatitis, pancreatic pseudocysts, and pancreatic tumors. This can also occur after a therapeutic endoscopy of the pancreas, including pancreatic sphincterotomy or stone removal. Hemosuccus pancreaticus can be difficult to detect by routine endoscopy; thus, a side-viewing duodenoscope and ERCP may be needed to reveal the source of the bleeding. The diagnosis may be confirmed by abdominal CT scan, angiography, or intra-abdominal exploration.

Bleeding occurs when there is erosion into a vessel, forming a direct communication with the pancreatic duct. Angiography is important for diagnosis and treatment, often with coil embolization to control the acute bleeding. For persistent or massive bleeding, surgery with resection of the bleeding area or ligation of a bleeding vessel may be required.

Cameron lesion

Diagnosis: Cameron lesions are erosions or ulcers that occur in the distal aspect of a hiatal hernia. Although this may be an incidental finding on upper endoscopy, these lesions can be responsible for iron deficiency anemia or acute or chronic upper GI bleeding.

The mechanism of formation of a Cameron lesion is not well understood. Potential etiologies include gastroesophageal reflux and mechanical trauma of the area. Management of a Cameron lesion depends on the clinical situation. Patients with acute bleeding can be treated by endoscopic methods. Patients who do not have acute bleeding, but have chronic blood loss may be treated with medical therapy, such as a proton-pump inhibitor and iron repletion. Surgical repair of the hiatal hernia is curative, but rarely necessary.

Upper gastrointestinal tumors

Diagnosis: Neoplasms of the upper GI tract are a rare cause of upper GI bleeding. These tumors may be esophageal squamous carcinomas, malignant lymphomas or adenocarcinomas (primary or metastatic often from lung or breast), or benign lesions such as leiomyeiomas or gastrointestinal stromal tumors. The symptoms that suggest bleeding may be from a gastrointestinal tumor include symptoms that may be attributable to the primary tumor such as dysphagia from esophageal cancer, but symptoms may be nonspecific including cachexia, weight loss, and early satiety. If the diagnosis of a tumor is not previously known, it may be detected at the time of endoscopy and confirmed by brushings or biopsies. Endoscopic ultrasound may be needed to evaluate submucosal masses.

Bleeding from an upper gastrointestinal tumor may be from mucosal erosions or ulcerations, or from invasion of the tumor into an underlying vessel. Typically, endoscopic treatments are a temporizing measure prior to more definitive measures such as surgery because rebleeding will frequently occur after endoscopy. Medical therapy is often ineffective in this setting, although palliative measures may be provided including chemotherapy and radiation of the primary tumor. Patients who have bleeding due to an upper GI malignancy have a very poor prognosis. The majority of patients (>60%) will die within 3 months. However, patients who have a benign upper gastrointestinal tumor that bleeds will be cured by surgical resection.

VARICEAL BLEEDING

Gastroesophageal varices

Diagnosis: Varices are dilated venous collaterals that have a tendency to rupture and can bleed massively. Esophageal and gastric varices develop as a result of portal hypertension usually due to advanced liver disease and cirrhosis (Figures 156-5 and 156-6). Varices may be isolated to the stomach if they are due to splenic vein thrombosis, acute pancreatitis, or a pancreatic tumor. ­Esophageal and gastric varices may occur without any clinical symptoms.

Acute variceal bleeding is different from nonvariceal bleeding in many respects. Upper endoscopy is the primary diagnostic modality and allows for direct visualization of columns of esophageal varices. Endoscopy should be performed quickly (within 24 hours) for potential intervention if varices are suspected as the source of upper GI bleeding. Abdominal CT and ultrasound may show the presence of collateral veins. Portal vein angiography may also demonstrate the presence of collaterals or recanalization of the umbilical vein.

Variceal hemorrhage is responsible for one-third of all deaths due to cirrhosis. Mortality is between 15% and 50% for each bleeding episode, and 70% and 80% in those with continuous bleeding. Following cessation of active variceal bleeding, there is a high risk of recurrent bleeding within 6 weeks (>50%). The time of greatest risk is within the first 48 to 72 hours (90% of rebleeds will occur in this time period), and over half of all subsequent rebleeding episodes occur within the first 10 days. Recurrent variceal bleeding may occur in up to 70% of patients within 6 weeks of the index bleed if preventive measures are not performed.

Survival during the 6 weeks following the index bleed is directly related to rebleeding. Risk factors include age greater than 60 years, large varices, severe initial bleed (hemoglobin <8 g/dL on admission) and renal failure. Patients with acute variceal bleeding are typically treated with multiple modalities simultaneously, including medical therapies and endoscopic management. Over-resuscitation should be avoided as overly vigorous volume repletion has been associated with precipitation of further bleeding. The target hemoglobin goal during resuscitation should be 7 g/dL. While early intervention in the form of transfusions (for low hemoglobin) and intravenous fluids may be warranted in the setting of hypotension, patients who are suspected to have variceal bleeding should undergo endoscopy within 12 hours of presentation and after resuscitation. In addition, radiologic and surgical treatments may be needed. Endoscopic variceal band ligation is a commonly applied therapy. This is a method of placing elastic bands over varices, similar to the technique of banding hemorrhoids. Varices are suctioned into a banding device and the bands are released around the base of the varices. It is typically performed in the distal 5 cm of the esophagus.

Pathophysiology: Normal pressure in the portal vein is 5 to 10 mm Hg. An elevated portal venous pressure of greater than 10 mm Hg (due to obstruction) increases capillary pressure. The connection between the portal and systemic circulation may enlarge to allow blood to bypass the obstruction and pass directly into the systemic circulation. One of the known sites of such confluence is the distal esophagus. Studies have demonstrated the role of endothelin-1 and nitric oxide in the pathogenesis of portal hypertension and esophageal varices. Endothelin-1 is a powerful vasoconstrictor synthesized by sinusoidal endothelial cells, that has been implicated in the increased hepatic vascular resistance of cirrhosis and in the development of liver fibrosis. Nitric oxide is a vasodilator substance that is synthesized by sinusoidal endothelial cells. In the cirrhotic liver, the production of nitric oxide is decreased, and endothelial nitric oxide synthase activity and nitrite production by sinusoidal endothelial cells are reduced.

Clinical findings: Symptoms of variceal bleeding are nonspecific and include hematemesis, melena, and hematochezia. Patients may feel light-headed and dizzy; and those with severe liver disease may have hepatic encephalopathy, which may be the sole presenting feature in a cirrhotic patient who is bleeding from the upper GI tract. Other associated signs and symptoms are the clinical manifestations of cirrhosis including the laboratory abnormalities associated with cirrhosis (see Chapter 160 [Cirrhosis and its Complications]). The objective of upper endoscopy is to find and define the source of bleeding and to treat it. Varices are evaluated for signs of either active bleeding or markers of recent bleeding. These markers include large tortuous varices with red wale marks (longitudinal red streaks on varices that resemble red, corduroy wales), cherry-red spots (discrete cherry-red spots that are flat on a varix) and hemocystic spots (raised discrete red spots that overlie varices that appear as “blood blisters”). Platelet or fibrin plugs are white nipple-like projections that project from a varix indicative of a site with recent bleeding and with a very high risk of bleeding.

Prognosis: Variceal bleeding is a significant cause of rebleeding and mortality. Variceal bleeding will stop in approximately 50% of patients, but those who have continued bleeding have a mortality that approaches 80%. Patients have a high risk of rebleeding (up to 70%) until the gastroesophageal varices are obliterated. If a patient survives the initial bleeding episode, repeated courses of band ligation are performed approximately every 2 weeks until the varices are obliterated. This is combined with medical therapy using nonselective β-blockade to reduce portal pressures. Current data support the use of nonselective β-blockers for medium and large varices prophylactically and for any grade of varix found to bleed. The prognosis for patients with bleeding gastroesophageal varices is poor even with control of bleeding varices as this is indicative of progressive liver disease. Patients die from hepatic decompensation, rebleeding, infections, renal failure, and other complications. Bacterial infections occur in up to 20% of cirrhotics with GI bleeding and subsequently develop in an additional 50% of patients during hospitalization for variceal bleeds. These include spontaneous bacterial peritonitis, pneumonia, sepsis, and skin infections. The use of prophylactic antibiotics in patients with acute variceal bleeding has been shown to decrease the rates of subsequent infection, spontaneous bacterial peritonitis, bacteremia, and death. Data for the use of fluoroquinolones as the prophylactic antibiotic of choice is supported by frequent clinical use, but a growing resistance of organisms to this class must be considered in choosing an antibiotic and alternatives include third-generation cephalosporins.

Portal hypertensive gastropathy

Diagnosis: In patients with cirrhosis and portal hypertension, gastric mucosal blood flow is increased, leading to congestion and hyperemia of the stomach. Portal hypertensive gastropathy occurs when edema and capillary venous dilatation in the stomach causes friability. This may subsequently result in bleeding with rupture of ectatic vessels. The endoscopic appearance is a mosaic-like pattern of pink mucosa, with a characteristic “snakeskin” appearance. Overall, this is a rare cause of acute upper GI bleeding, although it may be a source of chronic blood loss in patients with cirrhosis.

Portal hypertensive gastropathy may develop after treatment of esophageal varices. A proposed mechanism following treatment of varices is increased backpressure into the stomach vasculature, leading to the development of gastric congestion. Thus, the goal of treatment of portal hypertensive gastropathy is to decrease portal pressures. The affected area is diffuse and pharmacologic agents, such as octreotide, may be used acutely to decrease blood flow. Nonselective β-blockers (nadolol and propranolol) are often given, with the dose increased and adjusted if side effects occur. Endoscopic treatments are ineffective in this disorder. Patients with uncontrolled bleeding may need transjugular intrahepatic portal systemic shunt therapy or a surgical shunt in order to reduce portal pressure. In patients with decompensated liver disease, liver transplantation is indicated.

DIAGNOSIS

Upper GI bleeding may be life threatening. Factors that portend a favorable or a poor prognosis in patients with nonvariceal upper GI bleeding may be used to appropriately triage and manage patients. The role of nasogastric (NG) tube aspiration is controversial, and we do not routinely support its use. NG tube aspirates may be falsely negative; false-negatives most commonly occur in patients with bleeding from duodenal ulcers due to spasm of the pylorus, and may occur in other conditions, including gastric ulcers and rarely esophageal varices (if the tube is positioned in a nondependent area of the stomach).

HISTORY TO AID DIAGNOSIS

Similar to utilizing laboratory values to aid in diagnosis, a comprehensive history may help to predict a source of bleeding and prognosticate outcomes. An alcoholic with cirrhosis who has not recently undergone variceal screening endoscopy and presents with a large volume upper GI bleed should be managed as a variceal bleed until proven otherwise, including administration of octreotide and prophylactic antibiotics, and undergo urgent endoscopy. A patient with no history of liver disease or physical stigmata of cirrhosis probably presents with a nonvariceal source of bleeding. A patient with an extensive NSAID use history most likely has peptic ulcer disease.

LABORATORY TESTING TO ESTABLISH GI BLEEDING SOURCE

Although laboratory values cannot predict a source of bleed, they help to define the clinical status of the bleeding patient. The BUN will be elevated out of proportion to the creatinine due to digestion of blood in most patients with upper GI bleeding. In variceal bleeding, abnormalities of the liver enzymes (ALT, AST) are seen in patients with active hepatocellular damage. Patients may have hyperbilirubinemia and poor synthetic function, with hypoalbuminemia and an elevated INR indicating impaired liver function. Patients with bone marrow suppression from alcohol may have pancytopenia or those with hypersplenism may have low platelet counts. In patients with liver failure, hypoglycemia may be detected. Low albumin states may predict worse outcomes from bleeding.

TRIAGE/HOSPITAL ADMISSION

RISK ASSESSMENT: CLINICAL PREDICTORS AND GUIDELINES

Clinical guidelines have been developed to help optimize the management of patients with nonvariceal upper GI bleeding. The aim of guidelines is to identify low-risk patients who can be discharged either directly from the emergency room or at an early stage of hospitalization, and to identify high-risk patients who will need more resources. Several such guidelines include both clinical data and information obtained at the time of endoscopy. Newer guidelines focus on clinical data to determine which patients bear increased mortality risk and therefore merit urgent endoscopy and/or ICU level care.

Three main prognostic scores for upper GI bleeding have been used clinically. The Rockall score (Table 156-2) is a postendoscopy scoring system used to predict rebleeding and mortality in patients with nonvariceal upper GI bleeding. In this scoring system, scores of zero to three are assigned to the factors of age, the presence of shock, comorbidity, diagnosis, and endoscopic stigmata. A low-risk patient has a score of two or less, and about 30% of patients will belong to this category. In this low-risk group, there was a 4.3% risk of rebleeding and 0.1% mortality. Patients with Rockall scores of three to five have intermediate rates of rebleeding and mortality (2.0%-7.9%), whereas patients with a score of six or greater have a high rebleeding and mortality rate (15.1%-39.1%). Limitations of this scoring system include that it was derived from a relatively small number of patients and the full score requires both clinical and endoscopic information to calculate.

A scoring system that only uses clinical information (and not endoscopic data) obtained at the time of presentation prior to endoscopy, developed by Blatchford and colleagues (Table 156-3), helps to determine urgency of endoscopy and need for intervention. The clinical information incorporated into this scoring system includes hemoglobin, BUN, heart rate, systolic blood pressure, the presence of syncope, melena, liver disease, and heart failure. A new scoring system that also only uses information available at the time of initial presentation is the AIMS65 score. This score includes five factors: albumin <3.0 g/dL, INR >1.5, altered mental status, systolic blood pressure <90 mm Hg, and age >65 years. In this score, each factor is assigned a score of 1 and high-risk patients have an AIMS65 score >1.

MANAGEMENT

INITIAL HOSPITAL TREATMENT

While the early goals of treatment for GI bleeding include identification of the bleeding source and facilitation of bleeding cessation, the primary task is stabilization of the patient. Patients need to have at least two large bore peripheral intravenous access catheters (18 gauge or larger). Supplemental oxygen should be administered routinely. Crystalloid fluids should be initially administered to maintain an adequate blood pressure. Patients who are not able to adequately protect their airways (including patients with ongoing, severe hematemesis) should be considered for elective endotracheal intubation.

Patients who are high risk, such as the elderly or those with active coronary artery disease, should be transfused to maintain hemoglobin above 10 g/dL. Young and healthy patients should be transfused to maintain hemoglobin of 7 g/dL. Any coagulopathy should be corrected if possible with transfusion of fresh frozen plasma (initially 1 unit, followed by additional units based on subsequent INR) or administration of vitamin K (preferably 10 mg orally if at all possible). In patients with a low platelet count (<50,000/mL), it is recommended to transfuse platelets with a target platelet count goal >50,000/mL, initially with one or two units of platelets, followed by additional units based on subsequent platelet count.

Medications

Medications useful for management of acute upper GI bleeding include proton-pump inhibitors and octreotide. Other medications including H₂ blockers have not been demonstrated to favorably influence the natural history of acute GI bleeding.

Proton-pump inhibitor therapy

Proton-pump inhibitors help stop bleeding and prevent rebleeding in patients with significant nonvariceal GI bleeding. Endoscopic therapy is considered the standard of care for patients with active bleeding and nonbleeding visible vessels. Patients with peptic ulcer and successful endoscopic therapy have less rebleeding with the addition of intravenous PPI therapy (Table 156-4). In patients initially treated in the emergency room with a bolus infusion of omeprazole followed by a continuous infusion, active bleeding lesions found at endoscopy decreased and clean-based ulcers at endoscopy increased. A recent review and meta-analysis suggests that outcomes with an intermittent PPI dosing strategy is comparable to the previously recommended bolus and continuous infusion PPI dosing for patients with high-risk bleeding ulcers that are treated endoscopically. Patients who remain stable for 72 hours may then transition to standard dose oral PPIs.

Octreotide

The somatostatin analogue octreotide can be used to treat patients with both variceal and nonvariceal upper GI bleeding. This medication is indicated in patients with variceal bleeding or acute upper GI bleeding thought secondary to varices, using a loading dose of 25 to 50 μg followed by an intravenous infusion of 25 to 50 μg/h. Octreotide may also be helpful in patients with nonvariceal bleeding. While a useful adjunctive treatment in patients with non-variceal upper GI bleeding, octreotide should not be prescribed routinely to patients with nonvariceal upper GI bleeding. For all suspected etiologies of upper GI bleeding, the use of octreotide should be considered in patients who have persistent bleeding on optimal medical management, including PPIs, who are poor surgical risks (such as those with multiple comorbidities already in the hospital).

Endoscopic treatments

Endoscopic therapy for upper GI bleeding has been demonstrated to yield significant improvement in hemostasis, number of units of blood transfused, number of emergency interventions, hospital length of stay, and hospital costs. Endoscopic therapy is the standard of care for patients with the high-risk stigmata of recent hemorrhage of active bleeding or a nonbleeding visible vessel. The management of adherent clots has been controversial with recent data suggesting a possible role for endoscopic therapy.

For many conditions including active bleeding, visible vessels, and esophageal varices, endoscopic techniques are the mainstay of treatment. The timing and occasion for endoscopic treatment often depends on the clinical presentation and/or findings at time of endoscopy. An upper endoscopy examination is indicated within 24 hours of presentation for patients with suspected nonvariceal upper GI bleeding. Conditions that warrant rapid triage and planned endoscopy include bleeding in the setting of known varices, persistent hypotension requiring pressors, and increasing transfusion requirement. However, hemodynamic stability, protection of the airway (if necessary), and platelets, FFP, or vitamin K administration when applicable may be necessary steps prior to endoscopic evaluation regardless of the urgency. In addition, data support endoscopic evaluation within 24 hours and do not support a need for more urgent endoscopy in most patients with nonvariceal upper GI bleeding.

Optimization of visualization

One of the challenges of managing patients with GI bleeding is visualization during endoscopy due to blood within the gastrointestinal tract. This problem may be overcome by the use of a variety of techniques or a combination of endoscopic techniques including the use of double-channel or large-channel endoscopes and vigorous irrigation.

In addition to endoscopic techniques, intravenous erythromycin (250 mg IV bolus or 3 mg/kg over 30 minutes) can be used for its prokinetic properties to increase gastric emptying and clear the stomach of blood. The erythromycin is given intravenously 30 to 120 minutes prior to endoscopy as a useful adjunctive treatment in patients with large gastrointestinal bleeds. It can be used either initially or after an endoscopy shows large amounts of blood remaining in the stomach with withdrawal of the endoscope and the use of erythromycin before proceeding again with endoscopy. Metoclopramide may also be used similarly, although there are less available data about its effectiveness in patients with upper GI bleeding.

Methods to control bleeding

The current endoscopic modalities to treat nonvariceal GI bleeding include the use of injection therapies (primarily with dilute epinephrine), contact thermal therapies including heater and bipolar probes, noncontact thermal methods (predominantly argon plasma coagulation), mechanical treatments including a variety of clips including larger over-the-scope clips and band-ligation techniques, and a combination of the above treatment modalities (typically, injection therapies combined with one of the other modalities) (Table 156-5). For nonvariceal upper GI bleeding, combination therapy or use of hemoclips improves outcomes and decreases recurrent bleeding. For esophageal variceal bleeding, band ligation to tamponade blood flow offers effective treatment and outcomes. Success with injection of glues in the treatment of gastric varices has been demonstrated. Decompression of portal pressures when possible is the optimal therapy.

While endoscopic injection sclerotherapy and banding ligation in conjunction with pharmacologic treatment are the primary methods for controlling acute variceal bleeding, tamponade with a Sengstaken-Blakemore tube may be indicated when these methods fail or before endoscopy if a rapid bleed occurs. The use of these tubes is indicated for acute bleeding from esophageal or gastric varices unresponsive to medical therapy, tears at the gastroesophageal junction, Mallory-Weiss tears unresponsive to medical therapy, or esophageal exclusion to manage distal esophageal perforation. The Sengstaken-Blakemore tube is absolutely contraindicated for patients in whom bleeding has stopped and in those with recent surgery of the gastroesophageal junction or known esophageal stricture.

CONSULTATION

Hospitalists should involve a gastroenterologist early in the course of upper GI bleeding. Emergent consultation is appropriate with rapid bleeds in the setting of hypotension, a very low hematocrit, known varices, or other instances when urgent endoscopy might be required. In most cases, however, patient stabilization is a crucial first step before gastroenterology intervention can be considered (Figure 156-7). A gastroenterology consultant can provide diagnostic and therapeutic advice early in the course of treatment and may be called as soon as a suspected upper GI bleeding patient is identified. In addition, IR angiography and surgery consultations are indicated in certain circumstances including patients requiring the ICU, those who require multiple blood transfusions or those who undergo endoscopy without identification of a source of bleed or ability to stop the bleeding. In small bowel or lower GI bleeding patients, the patient may benefit from a tagged red blood cell scan and subsequent angiography. The key to consultation is to involve the consultant early and to communicate any significant clinical changes.

COMPLICATIONS AND PROGNOSIS

COMPLICATIONS OF GI BLEEDING

The reported mortality of upper GI bleeding generally ranges from 2% to 14%. Complications of GI bleeding include hypotension and subsequent shock, which can lead to organ failure and death if not addressed in a timely manner. Aspiration of blood contents and subsequent hypoxemia and respiratory failure represents another adverse outcome. Tachycardia in the setting of pre-existing coronary heart disease can lead to ischemia. Anemia can also further lead to underperfusion of vital organs and some patients die of complications subsequent to inadequate resuscitation with blood and fluids as discussed above.

COMPLICATIONS OF TREATMENT

The majority of the complications associated with treatment of upper GI bleeding stem from endoscopic treatments. Major complications include myocardial infarction, perforation, aspiration, hemorrhage, and death. Minor complications of endoscopy include mucosal tears, medication reactions, and hypoxemia. In addition, a small risk exists of infection or bleeding from the endoscopic intervention itself. As sedatives are utilized for the procedure, the individual risks of each medication are added to the overall risk of the procedure. Many of the complications of GI endoscopy are cardiopulmonary in origin.

The initial evaluation of patients with upper GI bleeding should include risk assessment to triage patients to appropriate levels of care and perform resuscitation measures. Elective endotracheal intubation should be considered in patients at high risk of aspiration, such as those with massive upper GI bleeding. Attention must be given to provide the proper level of sedation required to perform the procedure.

Epinephrine is a vasoconstrictor that can lead to local ischemia and can also cause tachycardia. Large doses of injected epinephrine during endoscopic therapy especially in the region of the esophagus and upper stomach are associated with significant elevations of systolic blood pressure and tachycardia. The use of cautery can lead to direct tissue injury, which may further damage the GI tract. The use of cautery may lead to perforation, especially if excessive pressure is exerted on the site of therapy. The depth of penetration of tissue injury may be lessened by the use of injection therapy that lifts the tissue and provides a safety barrier.

PROGNOSIS

In patients presenting with upper GI bleeding, a number of clinical prognostic factors have been shown to be helpful in predicting a poor outcome. These include older age (age >60 years), the presence of shock, hematemesis and/or hematochezia, onset of bleeding in a patient hospitalized for other reasons, bleeding from malignancies of the upper GI tract or varices, increasing comorbid diseases, and patients with severe coagulopathies.

DISCHARGE PLANNING

RESTARTING ANTICOAGULATION IN UPPER GI BLEEDING

Patients who require anticoagulants (eg, warfarin) or platelet function inhibitors (eg, clopidogrel, aspirin) may develop signs of upper GI bleeding. Most GI bleeding episodes on antiplatelet agents or anticoagulants occur within the first year of therapy. Once a source of bleeding is identified, the timing of restarting anticoagulants represents a difficult decision. Part of the decision-making process is evaluating the benefit of the anticoagulant or platelet inhibitor compared to the risk of bleeding. A patient with high-risk thromboembolic disease should have careful consideration of reversal of anticoagulation in consultation with the prescribing cardiologist or neurologist before endoscopy. A patient with low-risk thromboembolic disease can have complete reversal of anticoagulation followed by endoscopy. When to resume anticoagulation depends on the source and severity of bleeding as well as the underlying condition being treated. After hemostasis is achieved, patients who need to be restarted on anticoagulants have to be monitored carefully for subsequent bleeding.

OUTPATIENT SYMPTOM MONITORING

Patients discharged after upper GI bleeding require education about signs and symptoms to help identify any early bleeding. Patients should look for the return of black stool as a sign of a developing bleed. However, for the first 24 to 48 hours following active bleeding, patients can expect to see subsequent passage of blood, which may be maroon or black. Resting heart rate with evaluation for tachycardia can be measured by patients themselves. In addition, patients require education about over-the-counter medications to avoid that may affect GI bleeding, including drugs that contain NSAIDs, aspirin, or SSRIs. Bismuth subsalicylate (eg, Pepto-Bismol) should also be avoided as it causes the stool to appear dark and the stool color can be confused with melena.

OUTPATIENT LAB MONITORING

Patients who had bleeding related to supratherapeutic levels of warfarin need careful monitoring of the INR if the medication is subsequently resumed. Clinics that specialize in anticoagulation can reduce bleeding outcomes. In addition, anemic patients should have hemoglobin levels checked in the outpatient setting 1 to 2 weeks following discharge, as asymptomatic recurrent bleeding can present with worsening anemia. For upper GI bleeding, an elevated BUN can be a useful marker of recurrent bleeding. Patients with H. Pylori–related bleeding should have an H. pylori breath test or stool antigen performed if upper endoscopy is not scheduled to confirm that H. pylori has been successfully eradicated.

QUALITY IMPROVEMENT TO ADDRESS PERFORMANCE GAPS

PREVENTION

Primary and secondary prevention of upper GI bleeding requires careful evaluation of a patient’s underlying medical problems and current medications. NSAID use should be minimized whenever possible and, if necessary, one should monitor frequently for symptoms and signs of GI bleeding, which may indicate peptic ulcer disease, esophagitis, gastritis, or duodenitis, which can follow NSAID use. β-blocker therapy with propranolol or nadolol should be used in patients with cirrhosis and a prior history of esophageal variceal bleeding (secondary prophylaxis) regardless of grade, or large esophageal varices without previous hemorrhage (primary prophylaxis). The comparison of nonselective β-blockers and placebo in several randomized controlled trials demonstrates a greater than 20% absolute risk reduction in subsequent bleeding from esophageal varices. Prevention of GI bleeding in admitted patients requires careful monitoring of medications (eg, anticoagulants and antiplatelet agents) and comorbidities. Patients at high risk for stress ulcer bleeding (eg, mechanical ventilation) benefit from prophylactic PPI dosing.

TRANSITIONS OF CARE

As patients transition from in-hospital care to outpatient management, several factors must be considered. Medications must be adjusted from the inpatient setting to appropriate outpatient regimens. For example, although patients with upper GI bleeding are often managed with high-dose proton-pump inhibitors in the hospital, at discharge, most patients can be managed on a once daily standard dose oral proton-pump inhibitor or twice daily for high-risk bleeding lesions.

Findings at endoscopy may require changes in the medical regimen or subsequent endoscopies. In patients with peptic ulcer disease, biopsy results positive for H. pylori infection indicate a necessary outpatient antibiotic regimen. Most patients with significant gastric ulcers will require repeat endoscopy examinations in 8 to 12 weeks to confirm healing of the ulcer and for biopsy to exclude malignancy.

Communication of in-hospital caregivers with outpatient physicians is critical to the successful management of patients with significant upper GI bleeding. Follow-up with a primary physician should take place for repeat blood draws and physical examination within 7 to 14 days of hospital discharge, and follow-up with a gastroenterologist either for a planned repeat endoscopy or office visit should take place within 2 to 4 weeks depending on the source of bleeding. For example, variceal bleeding may require a 2-week ­follow-up endoscopy for further banding and assessment of healing. Outpatient planning for follow-up with gastroenterology, general surgery, and primary care should take place prior to discharge with the consulting specialists and the outpatient primary care provider.

DISPARITIES IN HEALTH CARE

While scarce literature identifies disparities in health care in gastroenterology, one recent study showed that significant proportion of visits to US emergency departments for acute GI illnesses are associated with a delay in initial clinical assessment. Of the patient visits analyzed, there were an estimated 1.6 million emergency department visits for acute pancreatitis, 2.2 million visits for appendicitis, 1.2 million visits for cholecystitis, and 3.9 million visits for upper GI bleeding. The study showed that Hispanic patients waited longer and had a higher frequency of delays compared with other racial and ethnic groups. Future investigations into racial, ethnic, gender and other disparities in health delivery and performance in upper GI bleeding are needed.

SUGGESTED READINGS

INTRODUCTION

EPIDEMIOLOGY

Acute pancreatitis is a common cause for hospitalization in the United States. Recent data from the National Center for Health Statistics have indicated a rising frequency of admissions for acute pancreatitis attributed to the increase in gallstone-related disease. Annually, more than 300,000 hospital admissions for acute pancreatitis present in the United States at a direct cost of more than $2 billion. Although acute pancreatitis is typically a mild, self-limited disease, a wide range of severity exists. Fifteen to twenty percent of patients experience a more severe form of disease with overall ­in-hospital mortality estimated to range between 2% and 5% of cases.

Risk factors for acute pancreatitis include both genetic susceptibility and environmental exposures. Genes involving mutation in the cystic fibrosis transmembrane conductance regulator, cationic trypsinogen gene (PRSS1), and secretory trypsin inhibitor (SPINK1) have been identified in patients with recurrent acute pancreatitis. Other patient-related factors such as obesity have also been demonstrated to be associated with increased severity of disease.

The most common etiology for acute pancreatitis is now ­gallstone-related disease, followed by alcohol and idiopathic pancreatitis (Table 157-1). ­Additional etiologies include metabolic derangements (­hypercalcemia, hypertriglyceridemia) and medications (Table 157-2). Less common etiologies include autoimmune and hereditary forms of pancreatitis. Whether pancreas divisum itself is a cause for acute pancreatitis remains controversial.

Obstructive causes for pancreatitis secondary to pancreatic mass or cystic lesions, such as intraductal papillary mucinous neoplasms are increasingly recognized due to improvements in imaging techniques.

RISK STRATIFICATION

Given the wide variation in morbidity and mortality in acute pancreatitis, risk stratification is essential. In 2012, the Atlanta ­Classification was revised and stratifies patients as mild, moderately severe, and severe pancreatitis. Mild acute pancreatitis is characterized by absence of organ failure and local or systemic complications, moderately severe acute pancreatitis is characterized by transient organ failure and/or local and/or systemic complications, and severe acute pancreatitis is characterized by persistent organ failure (≥48 hours).

Prognostic scoring systems have also been developed. ­Systemic inflammatory response syndrome (SIRS) has been associated with severe course. Systemic inflammatory response syndrome is defined as the presence of two or more of the following: temperature less than 96.8°F or more than 100.4°F, heart rate greater than 90 beats/min, respiratory rate greater than 20 breaths/min or PaCO₂ less than 32 mm Hg, white Blood Cell count greater than 12,000 cells/mL or less than 4000 cells/mL. The bedside index of severity in acute pancreatitis (BISAP) assigns points for each of the following: blood urea nitrogen (BUN) >25, impaired mental status, SIRS, age > 60, and pleural effusion. The presence of three or more of these findings signals a substantially increased risk of mortality. Other stratification tools such as APACHE-II score ≥8, laboratory findings such as BUN >20 or rising, hematocrit >44% or rising, CRP level at 48 to 72 hours >150 mg/dL, and risk factor including BMI ≥ 30 are also associated with a more severe course. Recent guidelines recommend the use of risk stratification.

DISORDER: EVALUATION, INPATIENT MANAGEMENT

EVALUATION

Patients with acute pancreatitis usually complain of sudden onset unrelenting epigastric pain that is sharp in nature. This pain often radiates to the back. Nausea and vomiting are common as is anorexia. Patients may present with signs of volume depletion in the setting of decreased oral intake as well as potential third-space losses. Signs of systemic inflammation are common at the time of presentation including fever, tachycardia, and leukocytosis. Rarely, physical examination findings may include evidence of hemorrhagic pancreatitis such as flank (Grey-Turner sign) or periumbilical (Cullen sign) ecchymosis.

The presence of epigastric pain with an elevation in amylase/lipase ≥3 times normal is highly accurate for the diagnosis of acute pancreatitis. Additional considerations should include acute cholecystitis or ascending cholangitis (both of which may be concurrent with an attack of biliary pancreatitis). Less common considerations include a penetrating duodenal ulcer, bowel ischemia, renal colic, or diabetic ketoacidosis. Further potentially life-threatening conditions include a perforated viscus, bowel obstruction or myocardial infarction. Serum amylase may also be elevated in salivary disorders, ectopic pregnancy, or macroamylasemia.

The diagnosis of acute pancreatitis requires at least two of the following: (1) typical epigastric abdominal pain, (2) amylase/lipase evaluation ≥3 times upper limit of normal, and/or (3) confirmatory findings on cross-sectional imaging. Findings on abdominal ultrasound in conjunction with liver function tests have very high sensitivity for the detection of gallstone-related illness. Specifically, an elevated ALT level > 2 to 3 times normal has been found to have a 95% positive predictive value for gallstone-related illness. More recently, magnetic resonance cholangiopancreatography (MRCP) has emerged as a highly accurate, noninvasive means to evaluate for the presence of biliary pancreatitis (Figure 157-1).

The finding of hypercalcemia in the setting of acute pancreatitis should prompt a thorough evaluation for the potential underlying etiology such as hyperparathyroidism or occult malignancy.

Autoimmune pancreatitis is a rare cause of acute pancreatitis that has recently gained increasing attention in the medical literature. An otherwise unexplained episode of acute pancreatitis in a patient with computed tomography (CT) scan findings of an enhancing “rim” or diffusely enlarged pancreas or pancreatogram features of an irregularly narrowed pancreatic duct are suggestive of autoimmune pancreatitis on imaging. A serum IgG4 level greater than two times the upper limit of normal may be useful to establish the diagnosis in type 1 autoimmune pancreatitis, which rarely presents as acute pancreatitis. Type 2 autoimmune pancreatitis, which is IgG4 negative, presents in 18% to 31% of cases as acute pancreatitis.

INPATIENT MANAGEMENT

Patients with confirmed acute pancreatitis should be managed as inpatients because they need vigorous intravenous fluid resuscitation and intravenous analgesia. A possible exception includes patients with a Harmless Acute Pancreatitis Score, which is defined as absence of rebound tenderness or guarding, normal serum creatinine, and normal hematocrit on presentation. These patients may be considered for a short hospital stay of less than 24 hours.

Persistent or worsening organ dysfunction after initial fluid resuscitation or patients with significant comorbid or concurrent illness (eg, congestive heart failure and/or chronic kidney disease) should be considered for hemodynamic monitoring in an intensive care setting. Also, patients with a rise in either hematocrit or BUN after initial resuscitation efforts should strongly be considered for hemodynamic monitoring to guide further resuscitation efforts.

Persistent multiorgan dysfunction and infection are two major complications that can develop in the patient hospitalized for acute pancreatitis. Patients with organ dysfunction are best managed in an intensive care setting. In addition to extrapancreatic infection, infected necrosis and persistent multiorgan failure are associated with a high morbidity and mortality. These patients are best managed in a multidisciplinary setting involving gastroenterology, surgery, and interventional radiology services.

Initial management of acute pancreatitis should include determination of etiology, assessment of severity, fluid challenge, and analgesia (Figure 157-2). Etiology may guide subsequent management, as in the case of gallstone pancreatitis. Clinical practice guidelines recommend abdominal ultrasound in the early assessment of patients with acute pancreatitis. Further assessment should include routine laboratory tests (liver profile, calcium, complete blood count). The use of CT scan of the abdomen in the initial 48 to 72 hours should be restricted to if the diagnosis of acute pancreatitis is uncertain. Fluid resuscitation is the cornerstone of initial treatment for acute pancreatitis. Clinical practice guidelines recommend vigorous initial fluid resuscitation with the thought of potentially preventing complications such as necrosis. Although aggressive fluid resuscitation has not been demonstrated to directly prevent specific complications, a rise in hematocrit or BUN in the setting of intravenous hydration has been associated with necrosis and in-hospital mortality, respectively. Specifically, an elevated BUN at admission >25 mg/dL and/or a rise in BUN during the initial 24 hours in the setting of volume resuscitation indicate an increased risk for in-hospital mortality (Figure 157-3). Therefore, patients with a rise in either hematocrit or BUN during initial resuscitation efforts should strongly be considered for hemodynamic monitoring to guide further resuscitation efforts. Aggressive fluid resuscitation should be focused in the initial 12 to 24 hours of presentation; continued aggressive hydration at 48 and 72 hours may be detrimental and associated with higher rate of SIRS and organ failure.

Adequate analgesia is central to patient comfort and recovery. ­Limited data are available regarding treatment options for pain in acute pancreatitis. However, there is no evidence to suggest a contraindication to morphine-related compounds due to sphincter of oddi spasm.

Recent studies have called attention to extensive use of imaging associated with acute pancreatitis. In an attempt to standardize use of imaging services, the American College of Radiology has issued appropriateness criteria for imaging tests in acute pancreatitis. ­Current recommendations are for the use of abdominal ultrasound early in the disease course to evaluate for biliary-related complications. Computed tomography is generally reserved for patients with evidence of systemic inflammation (SIRS) after 48 hours of illness or signs of clinical deterioration. If suspicion for biliary disease is high then MRCP may be considered as a noninvasive alternative to ERCP. Recent data also recommend the judicious use of contrast-enhanced CT scans to reduce patient exposure to radiation.

Endoscopic retrograde cholangiopancreatography is indicated within 24 hours when there is suspicion for ascending cholangitis. It is also indicated electively in the setting of persistent biliary obstruction (stone visualized on imaging with persistent elevation in liver function tests), or in patients deemed poor candidates for cholecystectomy. Fine-needle aspiration of pancreatic necrosis is indicated in a patient with evidence of persistent systemic inflammation in the absence of alternative source of infection (generally after 7-10 days of illness). Prophylactic antibiotics for the prevention of infected pancreatic necrosis are currently not recommended based on results from several recent double-blind randomized controlled trials. Antibiotics are considered appropriate while awaiting culture results in a patient with evidence of clinical deterioration. However, most experts advocate tailoring or removing antibiotics once culture results are known.

Gastroenterology should be consulted when gallstone pancreatitis is suspected, in the setting of a patient with persistent (>48 hours duration) SIRS or in the setting of acute pancreatitis complicated by necrosis and/or organ failure. Surgical consultation is indicated in the setting of gallstone pancreatitis in order to arrange for appropriate timing of cholecystectomy. Recent data have indicated that routine ERCP prior to cholecystectomy is not indicated. Interventional radiological services can assist in the identification of infected necrosis through CT-guided fine-needle aspiration and Gram stain of necrosis in a patient with ongoing fever or systemic inflammation. A multidisciplinary team approach is indicated for the management of infected necrosis as care may need to be individualized. In general, a multidisciplinary care can help optimize use of procedures, antibiotics, and imaging studies.

In mild acute pancreatitis, oral nutrition with clear liquids or low fat diet may be initiated once pain is controlled with analgesic. A recent randomized controlled trial showed that even in severe pancreatitis, oral nutrition started at 72 hours after admission was safe compared to early enteral nutrition started at 24 hours and was not associated with increased rate of infection or death. Nutritional support services may be required for initiation of enteral or parenteral nutrition in a patient that is not anticipated to begin oral feeding within 7 days of hospitalization. In addition, both consultation with gastroenterology and interventional radiology may assist in placement of a nasojejunal catheter for purposes of enteral nutrition in a patient that is not anticipated to begin oral feeding within 7 days. Enteral route should always be favored and parenteral nutrition should be reserved for patients unable to tolerate enteral feeding, or if enteral nutrition is not available.

POSTACUTE CARE

During the convalescent or recovery phase from acute pancreatitis, attention should focus on prevention of future recurrence and management of long-term complications. Recent data from a randomized controlled trial have indicated that repeated alcohol counseling can reduce the recurrence of alcohol-associated pancreatitis when compared to a single, one-time intervention. In addition, cases of “idiopathic” pancreatitis should be further evaluated with either repeat abdominal ultrasound (which may detect abnormalities such as biliary sludge often missed in the context of acute illness) or secretin enhanced MRCP to evaluate for anatomical anomalies including intraductal papillary mucinous neoplasm and pancreas divisum. More invasive interventions such as endoscopic ultrasound or ERCP are generally reserved for patients with recurrent episodes of idiopathic acute pancreatitis.

Long-term sequelae that may arise in the setting of acute pancreatitis include both local and systemic complications. Local complications include development of fluid collections that may either comprise a pseudocyst (homogeneous fluid-filled collection) or a postnecrotic collection (heterogeneous cavity filled with fluid and necrotic debris) (Figure 157-4). The distinction can be important given that it may influence approach to treatment. In general, intervention in fluid collections that arise in the setting of acute pancreatitis is only indicated in the setting of symptoms or infection. A symptomatic or infected fluid collection following an episode of acute pancreatitis is an indication for consultation with gastroenterology and surgical services. Although far less common, an additional potentially serious local complication is development of a pseudoaneurysm. Hemorrhage from a pseudoaneurysm can be life threatening and is best managed by angiography with surgical treatment reserved for failure to achieve hemostasis by alternative measures.

Additional long-term systemic complications that can result from acute pancreatitis include both exocrine and endocrine insufficiency of the pancreas. Patients with symptoms of steatorrhea or signs of fat-soluble vitamin deficiency should be evaluated with a fecal elastase on a solid stool to screen for pancreatic exocrine insufficiency. If the diagnosis is unclear, referral to a pancreas center for direct pancreatic function testing may confirm the diagnosis. This procedure consists of endoscopic collection of duodenal secretions after injection of intravenous secretin, which stimulates the pancreas to produce a bicarbonate-rich fluid. Once the diagnosis of exocrine insufficiency is confirmed, treatment with pancreatic enzyme supplementation is indicated (generally at least 48,000 IU lipase with each meal). Patients suffering necrotizing pancreatitis and especially those who have undergone pancreatic debridement are at increased risk of worsening or new-onset diabetes. Fasting glucose or hemoglobin A1C can be useful to screen for development of diabetes in this high-risk category of patients. Patients may develop brittle diabetes in the setting of extensive necrosis in which case referral to endocrinology may be useful in order to optimize glycemic control.

DISCHARGE CHECKLIST

Goals of discharge planning for acute pancreatitis include (1) arrangement for adequate follow-up, (2) prevention of disease recurrence, and (3) initiation of appropriate secondary evaluation. Although follow-up interval should be based on an individual patient’s needs, a 2-week interval has been routinely utilized. Patients with severe disease complicated by necrosis or fluid collections should be referred for outpatient follow-up with gastroenterology services. The two most important interventions for secondary prevention are cholecystectomy for gallstone-related acute pancreatitis and alcohol and smoking cessation counseling for patients with alcohol-associated disease. In the case of mild gallstone pancreatitis, cholecystectomy is strongly recommended during the same hospitalization to reduce the likelihood of recurrent episode of biliary complications while awaiting surgery. Prior studies have also shown this to be a cost-effective strategy for the prevention of gallstone pancreatitis. In the case of alcohol-associated disease, longitudinal alcohol cessation therapy is recommended.

SUGGESTED READINGS

INTRODUCTION

This chapter describes clinical conditions manifested as a result of acute or subacute biliary disease. Particular attention is paid to those conditions that frequently present to the inpatient setting, including jaundice, obstruction, and acute cholangitis.

JAUNDICE

Jaundice, from the French jaune, refers to the yellowish discoloration of the skin, sclera, and mucus membranes that reflects deposition of bilirubin in tissues. While jaundice is often used interchangeably with hyperbilirubinemia, this discoloration is only apparent when the serum bilirubin level exceeds at least two times the upper limit of normal (0.5-1.0 mg/dL), or over 2 mg/dL. Jaundice is typically recognized earliest in the oral mucus membranes (under the tongue and hard palate) and in the sclera, the latter of which has a high elastin content and a strong binding affinity for bilirubin.

PATHOPHYSIOLOGY

Jaundice results from a pathologic process that interferes with the normal metabolism and excretion of bilirubin (Figure 158-1). ­Physiologically, bilirubin is the product of hemoglobin breakdown formed during the elimination of senescent red blood cells. Macrophages, a component of the reticuloendothelial system, separate hemoglobin molecules into proteins (globulin) and heme. Splitting of the four pyrrole ring of heme produces biliverdin, which is then reduced to bilirubin. This type of bilirubin is unconjugated, and insoluble in water. As a result, it must be attached to albumin for transport to the liver for excretion.

Once in the liver, the bilirubin is rendered more water soluble by conjugation to glucuronic acid. Conjugated bilirubin is excreted into bile and subsequently enters the small intestine and colon. Exposure to oxidizing enzymes in the gut leads to the greenish color of bile. A portion of the bilirubin is deconjugated and reabsorbed into the circulation. The remaining segment is broken down by bacteria into urobilinogen, a fraction of which is converted into stercobilinogen and excreted in the stool, and the rest of which is excreted in the urine. Obstruction of the biliary system can hinder excretion of bilirubin into the gut and simultaneously exceed the processing ability of the kidneys, leading to pale stools and dark urine.

The amount of bilirubin in the serum is determined through measurement of light absorbance of the specimen following chemical treatment. Prior to examination, these specimens must be shielded from light to prevent excessive breakdown and falsely low results. The quantification of bilirubin is reported as direct (ie, conjugated), indirect (ie, unconjugated), and total (sum of conjugated and unconjugated forms). An increase in conjugated bilirubin is often the result of intra- or extrahepatic cholestasis. Elevations in unconjugated bilirubin, on the other hand, can occur as the result of three different pathophysiologic mechanisms including overproduction of bilirubin, impaired bilirubin uptake by the liver, or abnormalities of bilirubin conjugation.

CAUSES OF JAUNDICE

Prehepatic

Prehepatic jaundice occurs as a result of a pathologic process occurring outside the liver. The most common cause of prehepatic jaundice is hemolysis, whereby red blood cells are destroyed at an increased rate, thereby leading to excessive breakdown of hemoglobin. Fractionation of bilirubin in the serum of patients with prehepatic jaundice will typically show a predominance of unconjugated, or indirect, bilirubin. Other laboratory findings suggestive of hemolysis include a decreased serum haptoglobin level and an elevated peripheral blood reticulocyte count.

Causes of hemolysis are subdivided into those that are intrinsic (intracorpuscular) and those that are extrinsic to the red blood cell. The majority of intrinsic causes are inherited, and typically involve a defect in one or more of the red blood cell components including the hemoglobin protein, the red blood cell membrane, or the metabolic machinery required to generate adenosine triphosphate (ATP). Examples of these abnormalities include defective production of either alpha or beta globin chains in thalassemia, unstable or missing red blood cell membrane proteins as in spherocytosis, and defective defense mechanisms against oxidant challenge as in glucose-6-phosphate dehydrogenase (G6PD) deficiency, the latter of which is the most common genetic deficiency in the world.

Extracorpuscular causes of hemolysis are largely acquired and occur in a variety of contexts. Antibodies directed against the membrane components of red blood cells can lead to hemolysis, and occur in conditions such as autoimmune hemolytic anemia, transfusion reactions, and drug-induced processes. The mechanisms of and predispositions to drug-related liver injury are complex. However, many of these reactions result in drug-cholestasis, which is often reversible when the drug is withdrawn.

Infection may also cause extrinsic hemolysis through a variety of mechanisms including direct parasitization as seen in malaria and bartonellosis, damage from the release of bacterial products as seen in Clostridium and Haemophilus influenzae type b, and the induction of hypersplenism that leads to trapping and subsequent destruction of red blood cells, as seen in malaria. Malfunctioning cardiac hardware or disseminated intravascular coagulation (DIC) cause hemolysis, similarly leading to an indirect hyperbilirubinemia.

Hepatic

Intrahepatic mechanisms of jaundice result from an abnormality in the metabolism of bilirubin by the liver including problems with uptake from the circulation, intracellular storage, conjugation with glucuronic acid, or biliary excretion. The type of hyperbilirubinemia that arises with hepatic jaundice depends on whether or not conjugation has occurred, and therefore can present as an elevation in either direct or indirect bilirubin. Uptake of unconjugated bilirubin by liver cells is partly due to passive diffusion, and partly requires transport proteins. If the liver is unable to take up the unconjugated bilirubin for processing, levels will increase in the plasma. ­Furthermore, various forms of intrinsic liver disease and genetic polymorphisms can impair the liver’s ability to conjugate bilirubin, or alter the ability to subsequently secrete it into the bile. An example of this is in neonates, where immaturity of the liver may lead to jaundice because conjugation pathways are not fully functional.

Viral hepatitis, alcoholic hepatitis, drug or chemical toxicities, and autoimmune disease can lead to hepatocyte swelling or necrosis, which ultimately interferes with the aforementioned steps of bilirubin metabolism and results in jaundice. Replacement of normal liver parenchyma by either primary malignancy or metastatic disease can also cause jaundice by compressing normal adjacent tissue or bile canaliculi.

In the absence of other indicators of liver disease, genetic factors should be considered as a cause of jaundice. The most common inherited disorder of bilirubin glucuronidation, Gilbert syndrome, results from a defect in the promoter of the gene that encodes the enzyme uridine diphosphoglucuronate-glucuronosyl transferase 1A1 (UGT1A1), which is responsible for the conjugation of bilirubin with glucuronic acid. With the exception of intermittent episodes of jaundice, most patients with Gilbert syndrome are asymptomatic. Laboratory testing reveals unconjugated hyperbilirubinemia, with total bilirubin levels that are usually less than 3 mg/dL.

Crigler-Najjar syndrome is a rare, autosomal recessive disorder of bilirubin metabolism. It is due to a defect in the gene that encodes UGT1A1. Unlike Gilbert’s syndrome, which is the result of a defect in the promoter of this gene, these mutations lead to the production of an abnormal protein, resulting in complete loss or very low levels of UGT1A1. The phenotype can be severe jaundice and neurologic impairment due to bilirubin encephalopathy that can result in permanent neurologic sequelae (kernicterus).

Posthepatic

Several of the aforementioned disease processes may also contribute to posthepatic bilirubin elevations through inflammation or fibrosis of the intrahepatic bile ducts or canaliculi. As a result, these entities can potentiate obstruction of bile flow, also known as cholestasis. Extrahepatic mechanical obstruction to bile flow similarly causes a posthepatic obstruction and resultant jaundice, and is often the consequence of stones in the bile duct (choledocholithiasis), benign biliary strictures (eg, postsurgical, those associated with chronic pancreatitis, primary sclerosing cholangitis [PSC]), infiltrating tumors, or compression from adjacent carcinomas such as malignancy involving the head of the pancreas. Rarer causes include pancreatic pseudocysts, bile duct parasites, biliary varices, and inflammatory pseudotumor. As a result of backflow of conjugated bilirubin, serum direct bilirubin concentration increases in obstructive processes.

BILIARY OBSTRUCTION AND ACUTE CHOLANGITIS

Biliary obstruction refers to the blockage of any duct that carries bile from the liver to the gallbladder or from the gallbladder to the small intestine. Intrahepatic cholestasis generally occurs at the level of the hepatocyte or biliary canalicular membrane. Extrahepatic obstruction to the flow of bile, on the other hand, may occur within the ducts or secondary to external compression.

As detailed above, a variety of conditions can lead to obstruction of bile flow. In the setting of biliary obstruction, intrabiliary pressure rises, leading to increased permeability of the bile ductules and predisposing to translocation of bacteria from the portal circulation into the biliary tract. When the biliovenous pressure exceeds about 40 mm Hg, bacteria can migrate into the bloodstream. This increase in pressure also alters a variety of host-defense mechanisms including Kupffer cells, bile flow, and IgA production. With disruption of these barrier mechanisms, biliary bacteria are able to access the venous system.

Biliary obstruction and stasis are the most important predisposing factors leading to the development of biliary infection, known as acute cholangitis. In developed countries, the common causes of acute cholangitis include bile duct stones, which can be subdivided into primary (eg, intrahepatic stones), secondary (from gallbladder), and complicated stones (eg, Mirizzi syndrome), in addition to benign biliary strictures (eg, postsurgical, chronic pancreatitis, primary sclerosing cholangitis) and obstructed biliary stents and drains (catheters). Malignant biliary strictures from cholangiocarcinoma, and cancers of the gallbladder, ampulla and pancreas, less commonly present with acute cholangitis, but if unrelieved, can lead to biliary sepsis and liver abscesses (Figure 158-2). Parasites (eg, Clonorchis sinesis, Fasciola hepatica), viruses (eg, AIDS cholangiopathy), and fungal infections (eg, candida), may cause biliary obstruction and cholangitis. Additionally, interventions such as endoscopic retrograde cholagniopancreatography (ERCP), percutaneous transhepatic cholangiography (PTC) and surgery may result in the loss of the physiologic barrier between the bile duct and intestine, and facilitate entry of bacteria into the obstructed biliary system.

Inciting organisms typically ascend from the duodenum, although hematogenous spread from the portal vein has also been reported. These organisms are typically Gram-negative bacteria including Escherichia coli, Klebsiella, and Enterobacter, although infection with Enterococci, a Gram-positive bacterial species, is not uncommon. Unrelieved, acute cholangitis carries close to 100% mortality. Even when the biliary tree is decompressed, the mortality of acute cholangitis is not insignificant.

PRESENTATION AND DIAGNOSIS

Physical findings

The major signs and symptoms of biliary obstruction result directly from the failure of bile to reach its proper destination. Acute cholangitis, also referred to as ascending cholangitis, is characterized by a clinical syndrome known as Charcot’s triad, which consists of pain, jaundice, and fever. The addition of hypotension and confusion to Charcot’s triad suggests sepsis, and the combination of five symptoms makes up Reynolds Pentad. Diaphoresis and tachycardia also commonly occur, although these may be absent in elderly patients who may instead present with confusion or hypotension as the only sign.

Since Charcot reported a patient with severe acute cholangitis as a case of “hepatic fever” in 1877, Charcot’s triad has been widely regarded as one of the most important diagnostic criteria. However, despite its high specificity, it has extremely low sensitivity. Moreover, only half to three-quarters of patients with acute cholangitis have all three findings of Charcot’s triad. As a general rule, the diagnosis of acute cholangitis should be suspected if a patient has fever or shaking chills and laboratory evidence of an inflammatory response, in addition to either jaundice or abnormal liver function tests. Biliary dilation on imaging or evidence of an etiology such as a stone, stricture, or stent that is visualized confirms the diagnosis.

The most common presenting symptoms of acute cholangitis are fever and abdominal pain. Typically, the abdominal pain is either localized to the right upper abdomen or more diffuse. Surgical signs, such as abdominal rigidity and rebound tenderness, are absent, unless there is concurrent acute cholecystitis. Jaundice is a less common presenting sign, but when it is present, it tends to first occur in the sclera of the eyes or sublingually.

Unless a patient has concurrent chronic liver disease, he or she will not exhibit stigmata of chronic liver disease at the time of presentation, such as spider nevi, palmar erythema, or finger clubbing. Furthermore, the liver and spleen should not be enlarged, and no masses should be present. The one exception to this is palpation of an acutely distended gallbladder, which can occur in mechanical obstruction. This is rare in benign disease, but is the cause of ­Courvoiser sign in malignant obstructive jaundice.

Laboratory findings

Appropriate blood tests on presentation include a complete blood count, a comprehensive metabolic panel (including a liver panel), and basic coagulation studies. Typically, laboratory studies will reveal a leukocytosis with neutrophil predominance, in addition to a cholestatic pattern of liver test abnormalities including an elevation in both conjugated bilirubin and alkaline phosphatase.

Aerobic and anaerobic blood cultures should be drawn before the institution of antibiotics, and treatment modified as necessary when results are available. Bile is normally sterile; however, in the presence of gallbladder calculi or common bile duct (CBD) stones, the incidence of bactibilia increases.

Gram-negative bacteria cause the majority of cases of acute cholangitis. The most common organisms isolated in bile are E. coli (27%), Klebsiella species (16%), Enterococcus species (15%), Streptococcus species (8%), Enterobacter species (7%), and ­Pseudomonas aeruginosa (7%). Cultures should also be obtained from bile or stents removed at the time of endoscopic retrograde cholangiopancreatography. Organisms isolated from blood cultures are similar to those found in bile. The most common pathogens isolated in blood cultures are E. coli (59%), Klebsiella species (16%), P. aeruginosa (5%), and Enterococcus species (4%). In addition, polymicrobial infection is commonly found in bile cultures (30%-87%) and less frequent in blood cultures (6%-16%).

Imaging

Transabdominal ultrasonography may show dilatation of the common bile duct (normal diameter ≤8 mm), with or without the presence of stones in patients with choledocholithiasis (Figure 158-3). Biliary dilation greater than 8 mm in a patient with an intact gallbladder is usually indicative of biliary obstruction. A stone in the common bile duct seen on ultrasound is the most reliable predictor of choledocholithiasis at subsequent endoscopic retrograde cholangiopancreatography or surgery. Ultrasonography also provides useful information about the gallbladder in regard to cholelithiasis or cholecystitis.

Transabdominal ultrasonography may be falsely negative when only small stones (<6 mm) are present in the bile ducts or in patients with acute obstruction when the bile duct has not yet started to dilate. In this situation, magnetic resonance cholangiopancreatography (MRCP) is recommended to exonerate the bile duct of stones or other missed diagnoses (Figure 158-4). In patients who have a contraindication to MRI, ERCP is often performed if the clinical picture and laboratory studies suggest biliary obstruction. ­Cholangiography performed during ERCP may confirm the presence of biliary obstruction from choledocholithiasis, in addition to benign or malignant strictures (Figure 158-5). Endoscopic ultrasound (EUS) is also of great utility in patients who have more equivocal presentations for acute cholangitis or in whom ERCP is high risk. The use of EUS to determine if ERCP is indicated may avoid a significant number of ERCPs and result in fewer complications.

In early stages of acute gallstone pancreatitis, commonly used clinical predictors of biliary obstruction, such as cholestatic liver enzymes and imaging demonstrating a dilated CBD, are less reliable in predicting the presence of CBD stones. As a result, MRCP and/or EUS may be warranted in selecting patients for therapeutic ERCP. Furthermore, when the suspicion of acute gallstone pancreatitis is low to moderate, in equivocal presentations for acute cholangitis, and in patients who are high risk for ERCP, these alternative imaging modalities can be useful for exonerating the bile duct of stones or other missed diagnoses.

DIFFERENTIAL DIAGNOSIS

Patients with a variety of other disorders may also present with symptoms such as fever and abdominal pain, and a broad differential should be entertained in those with either atypical presentations or normal abdominal imaging. The differential ranges from biliary leaks to diverticulitis to liver abscesses to intestinal perforation, and even includes intrathoracic processes that can mimic right upper quadrant abdominal pain such as right lower lobe pneumonia or empyema.

MANAGEMENT

Initial hospital treatment

Acute cholangitis requires rapid relief of biliary obstruction by endoscopic, percutaneous, or surgical intervention. Patients who present with signs and symptoms of acute cholangitis should have liver and biliary imaging and basic blood work (complete blood count, comprehensive metabolic panel, blood cultures) performed without delay to support the diagnosis. Early initiation of antibiotics is also crucial in the management of patients with acute cholangitis.

Although there is no single accepted scoring system, the American Society for Gastrointestinal Endoscopy (ASGE) provides a proposed strategy to assign risk of choledocholithiasis in patients with symptomatic cholelithiasis. Very strong predictors of choledocholithiasis include a CBD stone on transabdominal ultrasonography, ascending cholangitis, or a bilirubin of over 4 mg/dL. Strong predictors include a dilated CBD on ultrasound or a bilirubin elevation of 1.8 to 4 mg/dL. Moderate predictors include abnormal liver biochemical tests other than bilirubin, age older than 55 years, or clinical gallstone pancreatitis. The probability of choledocholithiasis is categorized as high (>50%) when any of the very strong predictors or both of the strong predictors are present, low (<10%) when no predictors are present, and intermediate (10%-50%) for all other patients.

Typically, patients with a high probability of choledocholithiasis based on clinical presentation and noninvasive imaging are best served by proceeding directly to ERCP. Exceptions include patients in whom prior attempts at ERCP were unsuccessful or in whom the risk of pancreatitis and/or radiation exposure would make confirmatory testing desirable (eg, gallstone pancreatitis, pregnant women).

The majority of patients will respond favorably to antibiotic therapy. If a response to antibiotics is seen, it is reasonable to wait until arrangements can be made to perform an ERCP in the endoscopy suite with experienced staff, provided the procedure can be done within 24 to 48 hours. If ERCP is delayed and the patient has not improved over the first 24 hours with conservative management, urgent biliary decompression is required. In patients who develop signs of worsening disease or biliary sepsis (including hypotension despite adequate fluid resuscitation, fever greater than 102°F, or mental confusion), or in patients with acute gallstone pancreatitis and coexisting cholangitis, urgent biliary decompression is indicated given the observed benefits in morbidity and mortality.

Endoscopic sphincterotomy with stone extraction and/or stent insertion (depending on the cause of the obstruction) is the treatment of choice for establishing biliary drainage in acute cholangitis, and resolves nearly all episodes (Figure 158-6). In patients with large stones (>1.5-2 cm) or narrow distal bile ducts relative to the size of the stone, large diameter endoscopic balloon dilatation immediately after sphincterotomy reduces the need for lithotripsy techniques. In general, an impacted ampullary stone can make biliary deep cannulation difficult and necessitate the need for precut sphincterotomy.

Acute cholangitis should be regarded as a medical emergency and treated aggressively. If ERCP is not technically feasible or fails to adequately re-establish bile drainage, PTC or open surgical decompression are the next steps at achieving biliary drainage. PTC allows external access to the biliary tree. Once access has been obtained, a percutaneous drain can be inserted for decompression. If the obstruction can be traversed, an internal-external drain is placed, with the distal end left in the duodenum (though the duodenal papilla) (Figure 158-7). The drain is usually held in position at the skin access site by a suture. Initially, the bile is allowed to drain externally into a bag. After the sepsis has resolved, the drain is internalized (capped at skin level), so that bile flows into duodenum through the catheter. To maintain patency, the internalized drain is typically flushed with sterile water two or three times a day.

It is unusual for surgery to be performed to manage cholangitis alone. In the setting of acute cholecystitis with acute cholangitis, open cholecystectomy with common bile duct exploration may be performed, with simultaneous placement of a T-tube for biliary drainage. However, it is preferable to have preoperative biliary decompression via ERCP if obstruction and sepsis are present. If acute cholecystitis is identified complicating acute cholangitis, the placement of a cholecystostomy tube under radiologic guidance is a useful temporizing maneuver until resolution of the infectious cholangitis. Elective cholecystectomy may then be considered. Additionally, in postsurgical strictures that fail endoscopic or percutaneous stenting, as evidenced by progressive liver damage and/or recurrent episodes of acute cholangitis, surgical bypass (eg, choledocho-jejunostomy) can reduce recurrence and worsening organ dysfunction. Furthermore, in a small subset of patients with primary sclerosing cholangitis, liver transplantation may be necessary to manage progressive deterioration due to recurrent cholangitis and cirrhosis. Patients who have chronic biliary sepsis are generally considered unsuitable for liver transplantation, and therefore, aggressive treatment of acute cholangitis episodes is paramount in the management of their disease.

Medical management in acute cholangitis includes fluid resuscitation, correction of electrolyte abnormalities, correction of infection-related complications (eg, disseminated intravascular coagulation), and administration of antibiotics. Patients with severe sepsis or shock should be managed in an intensive care setting, with aggressive diagnostic and therapeutic measures including central venous catheterization and vasopressors when needed. While there is no consensus regarding the best initial antibiotic regimen for cholangits, patients are typically given broad-spectrum parenteral antibiotics aimed at colonic bacteria, taking into account local patterns of resistance. Regardless of initial drug regimen, therapy should be modified to reflect any organism(s) recovered in blood cultures.

PREVENTING RECURRENCE

Patients who develop acute cholangitis due to gallstones are at risk for recurrence. As with other complications of gallstone disease, cholecystectomy is generally recommended. If the obstruction is due to a benign stenosis, as is seen following bile duct injuries, endoscopic therapy or surgical repair may be required. Recurrent obstruction is common in patients with malignant stenosis. Management is typically with stent placement, though the specific therapy chosen will depend on the patient’s life expectancy and the likelihood of stent occlusion.

Furthermore, patients with recurrent cholangitis from primary sclerosing cholangitis may also benefit from cycling oral antibiotics, although no standard guidelines exist for antibiotic regimens to maintain bile sterility. One week on and 2 weeks off, oral ciprofloxacin 500 mg twice daily may be considered. Metronidazole may be added for anaerobic coverage, although it frequently causes a metallic taste and nausea. Length of cyclical therapy does not have supporting studies to guide practice. A trial of 3 to 6 months may be reasonable before a trial off of cycling antibiotics, and clinicians should monitor and record episodes of recurrence while on and off preventive antimicrobial therapy. Episodes of cholangitis while on oral antibiotic therapy suggest unrelieved biliary obstruction or antibiotic resistance, or both. A different antibiotic regimen should be substituted if there is recurrence or concern for resistance while on prophylactic therapy.

PROGNOSIS

Biliary sepsis should be regarded as a medical emergency and managed with appropriate urgency. Acute cholangitis without adequate biliary drainage is associated with mortality approaching 100%. Since the introduction of antibiotics effective against Gram-negative bacteria and nonsurgical means of decompressing the biliary tree (ERCP and percutaneous drainage), the mortality has decreased dramatically. However, severely ill patients with cholangitis who have major comorbidities still may not survive, even if effective biliary drainage can be re-established.

Unrelieved chronic biliary obstruction is often associated with a low-grade cholangitis, typically seen in the setting of strictures of the biliary tree. Eventually, the part of the liver affected by the obstruction may undergo atrophy or develop secondary biliary cirrhosis. Chronic obstruction may also lead to the development of multiple abscesses within the liver.

DISCHARGE PLANNING

Discharging clinicians should assure adequate biliary drainage for patients leaving the hospital. Patients who have undrained or inadequately drained bile ducts are likely to return to hospital with recurrent cholangitis, even if maintained on the appropriate antibiotic regimen. Occasionally, the most effective form of biliary drainage is external, especially in patients with complex malignant hilar strictures. These patients often leave the hospital with internal-external biliary catheters. Biliary drains need to be flushed with sterile water several times a day.Discharging physicians should arrange for home health nursing staff to assist with these drains and also teach patients how to care for these drains themselves.

Patients with recurrent cholangitis should be referred to and followed by a physician who understands the natural history of the disorder, including a gastroenterologist or surgeon with interest and expertise in the management of hepatobiliary disorders. Patients who may require surgery should be jointly managed by a gastroenterologist and a surgeon. Patients with chronic liver disease related to biliary obstruction and cholangitis should also be followed by a transplant hepatologist who will coordinate evaluation and listing for liver transplantation when indicated. Physicians managing patients on cyclical antibiotics often should consult infectious disease colleagues who will advise on the choice of antibacterial agents and their rotation, if needed. Patients at risk for recurrent cholangitis, who do not receive long-term antibiotics, should be provided with standing prescriptions for antibiotics, which they should start at the first sign of infection.

DISPARITIES IN HEALTH CARE

Patients who lack medical insurance frequently delay seeking care for symptoms of cholangitis, leading to increased mortality among this population. Indigent patients with comorbidities, such as chronic viral hepatitis or human immunodeficiency virus, have particular risk for overwhelming sepsis in the setting of cholangitis. The underinsured and uninsured typically lack outpatient health care providers in the community, and rely on emergency department care when they become seriously ill. Such patients should be identified when they present to health care facilities, so that social services and related resources are applied to find longitudinal care by appropriate physicians and other care providers.

OUTCOMES TO MONITOR

Patients with acute cholangitis do not always present with the classic symptoms described above. Rather, symptoms may be quite subtle, such as vague abdominal pain with or without shaking chills or a flu-like illness. This is especially true for patients with chronic cholangitis, and modified by intermittent courses of antibacterial agents. In some patients, the only symptom signaling biliary obstruction or cholangitis may be pruritus.

Patients at risk for recurrent attacks of cholangitis should be educated to recognize the early clinical signs, and encouraged to seek medical care as soon as they do. Cholangitis may be asymptomatic, and lead insidiously to chronic liver disease. One clue to asymptomatic cholangitis includes a progressive increase of liver alkaline phosphatase. When liver disease is established, decompensation accompanying biliary sepsis may present as confusion due to hepatic encephalopathy. Other complications of cirrhosis, especially bleeding from esophageal varices, may supervene. Regular clinic appointments and monitoring liver tests may help to identify those patients who have inadequate biliary drainage and are at risk for recurrent cholangitis and cirrhosis.

COSTS AND RESOURCE UTILIZATION

Early, aggressive management of acute cholangitis will help avoid costly complications, and reduce overall mortality. Tertiary care center referral for acute cholangitis is indicated if the admitting hospital is not equipped to perform ERCP or other necessary testing and interventions. Late transfer of a very ill patient to a specialist or referral center likely worsens morbidity and mortality.

CONCLUSION

The management of jaundice requires integration of bilirubin metabolism physiology with the pathologic processes that may interfere with this complex metabolic process (prehepatic, hepatic, and posthepatic mechanisms). Biliary obstruction and stasis are the most important predisposing factors leading to the development of biliary infection, known as acute cholangitis. This entity should be suspected if a patient has abdominal pain, fever or shaking chills, and laboratory evidence of an inflammatory response, in addition to either jaundice or abnormal liver function tests. Confirmation of the diagnosis is made when the patient also has biliary dilation on imaging or evidence of an etiology such as a stone, stricture, or stent that is visualized. Unrelieved biliary obstruction can be fatal. Early recognition of acute cholangitis and the prompt re-establishment of biliary drainage is the key to successful management. ERCP is the treatment of choice. However, if ERCP is not technically feasible or fails to adequately re-establish bile drainage, PTC or open surgical decompression are the next steps at achieving biliary drainage. As elderly and immunosuppressed patients often fail to exhibit the classic signs and symptoms of acute cholangitis, a high index of suspicion is necessary to make the diagnosis.

SUGGESTED READINGS

INTRODUCTION

DEFINITION

Acute liver failure (ALF) is a rare clinical syndrome that is defined by coagulopathy, usually an international normalized ratio (INR) >1.5, and any degree of mental alteration (encephalopathy) that occurs over a span of less than 26 weeks in a patient without pre-existing liver disease. Based on the time to encephalopathy from the onset of jaundice, three types of ALF have been described: hyperacute (0-7 days), acute (8-21 days), and subacute (>21 days and <26 weeks).

Patients with subacute liver failure often pose a clinical challenge, as they may show signs suggesting chronic liver disease, such as ascites or prolonged encephalopathy. Also, patients with subacute liver failure (eg, drug induced) may follow a variable and unpredictable course, often leading to heightened anxiety for the patient, family, and medical providers.

As an exception to the definition of ALF, patients with Wilson disease, reactivation/superinfection of chronic hepatitis B, or autoimmune hepatitis may be considered to have ALF despite the presence of underlying chronic liver disease.

EPIDEMIOLOGY

Approximately 2000 cases of ALF occur each year in the United States. ALF can affect people of all ages and is associated with high morbidity and mortality. Multiorgan failure (MOF) is the most common cause of death (>50% of cases), with infection and intracranial hypertension (ICH) responsible for most of the remaining deaths. With recent improvements in care, spontaneous recovery and overall survival now exceed 40% and 65%, respectively. Over the past 25 years in the United States, overall mortality from ALF has decreased from more than 80% to 33%.

PATHOPHYSIOLOGY

Acute liver failure involves massive hepatocellular necrosis and liver dysfunction. A resulting cytokine storm leads to oxidative stress and a systemic inflammatory response syndrome. Anti-retroviral medications, Wilson disease, and acute fatty liver of pregnancy may uniquely cause severe mitochondrial dysfunction with resulting lactic acidosis. Ultimately, multiorgan dysfunction occurs.

Intracranial hypertension and cerebral edema may result from elevated ammonia levels that cause direct astrocyte toxicity and loss of cerebral blood flow autoregulation. Loss of cerebral autoregulation in ALF leads to ICH, and intracranial contents are rapidly damaged by both hydrostatic and osmolar shifts. Noncardiogenic pulmonary edema (acute respiratory distress syndrome [ARDS]) and renal failure may result from cytokine-induced damage and hemodynamic derangements. Cardiovascular collapse can be seen with a severe depression of systemic vascular resistance, another process that is driven by cytokines and derangements in vasoactive substances. Loss of the normal synthetic function of the liver also leads to both a coagulopathy and prothrombotic state. Metabolism of medications and other physiologic mechanisms, such as biliary excretion, are also disrupted in the setting of ALF.

Acute liver failure differs from acute-on-chronic liver failure. Although there are some examples of overlap (eg, Wilson disease, autoimmune hepatitis, acute-on-chronic hepatitis B), the pathophysiology of the two clinical syndromes differ with respect to physiologic adaptation. For instance, patients with ALF are at risk for elevated intracranial pressure (ICP) because neuronal adaptation does not have time to occur as the ammonia levels rapidly rise.

DIFFERENTIAL DIAGNOSIS

In the setting of possible ALF, clinicians must exclude chronic liver disease. Only a few chronic hepatic conditions may present with true ALF: (1) Wilson disease, (2) autoimmune hepatitis, and (3) acute-on-chronic viral hepatitis B.

A nodular-appearing liver on imaging studies can be misleading, as an irregular contour may represent regenerative nodules interspersed with areas of acute injury and necrosis. Similar to acute alcoholic hepatitis, acute portal hypertension and ascites may also be seen with ALF. Finally, sepsis with disseminated intravascular coagulation in a patient who has cirrhosis can also mimic ALF, as such a patient can present with altered mental status, coagulopathy, and jaundice.

CAUSES OF ALF

In the United States, acetaminophen toxicity is the number one cause of ALF (39%). Approximately 20% to 25% of these patients undergo orthotopic liver transplantation. Whereas most cases of acetaminophen-induced ALF involve ingestions exceeding 10 g/d, severe liver injury can occur with doses as low as 3 to 4 g/d, as is seen with therapeutic misadventure. Therapeutic misadventure, referring to therapeutic doses of acetaminophen in the setting of alcohol, malnutrition, or cytochrome p450-inducing medications, occurs in nearly half of acetaminophen toxicity cases (Table 159-1).

In addition to acetaminophen, multiple other causes can lead to ALF, and in the United States, approximately 20% of ALF cases have no discernible cause (Table 159-2). Non-acetaminophen drug or toxin-induced liver injury accounts for 13% of cases (Table 159-3). Toxins may include Amanita phalloides (mushrooms), sea anemone sting, or carbon tetrachloride. Acute hepatitis A and B account for 4% and 7% of ALF cases, respectively. The remainder of cases may be due to other viruses (herpes simplex virus [HSV], Epstein-Barr virus [EBV], cytomegalovirus [CMV]), Wilson disease, Budd-Chiari syndrome, sinusoidal obstruction syndrome, acute fatty liver of pregnancy, autoimmune hepatitis, malignant infiltration, or ischemia.

DIAGNOSIS

PRESENTATION

A patient can present at any point along the disease severity spectrum. Multiorgan failure may rapidly ensue and manifest with abrupt clinical decompensation. Some patients, such as those who recently ingested an acetaminophen overdose, or those with subacute liver failure, may initially present asymptomatically, despite having markedly abnormal bloodwork. Nonspecific symptoms may include nausea, vomiting, abdominal pain, fever, malaise, fatigue, arthralgias, itching, bleeding, headache, and diarrhea. Nonspecific clinical signs may include abdominal pain, altered mental status, jaundice, or ascites.

HISTORY

A detailed history, including a comprehensive review of possible exposure to viruses, drugs, or toxins, may help uncover a possible cause of liver failure. Clinicians must question about over-the-counter medications, alternative therapies, herbs, and supplements (Table 159-3). If the patient is encephalopathic, a collateral history from friends or family is important, as identifying a cause can be paramount in the management of ALF.

Acetaminophen toxicity may involve intentional overdose or may involve high doses over an extended period of time. For example, a patient who takes over-the-counter acetaminophen in addition to prescribed acetaminophen-containing pain medications may exceed a critical daily dose that leads to severe liver injury. Alcoholics and malnourished patients are at greater risk for acetaminophen-induced liver injury.

Most idiosyncratic drug-induced liver injuries are not dose related and tend to occur within the first 6 months of drug initiation (Table 159-3).

A history of IV drug abuse may indicate risk for infections such as acute hepatitis B and D. Travel to endemic regions (eg, ­Russia, ­Pakistan, India, Mexico) may place a patient at risk for acute hepatitis E infection. Immunosuppressed patients and women in their third trimester of pregnancy are at increased risk for herpes simplex infection. A history of preeclampsia in a pregnant woman increases the risk of acute fatty liver of pregnancy, which also occurs in the third trimester. Previous or coexisting autoimmune diseases (­thyroiditis, insulin-dependent diabetes, lupus, vitiligo, etc) may indicate an increased risk for autoimmune hepatitis. Cardiac disease (congestive heart failure, myocardial infarction, arrhythmia) or hemodynamic instability may lead to “shock liver” from hypoperfusion. Documented hypotension is not always found with ischemic liver injury. A history of malignancy or hypercoagulable state may raise concern for malignant infiltration or Budd-Chiari syndrome (hepatic vein thrombosis), which often presents with acute abdominal pain, ascites, and striking hepatomegaly. A history of chemotherapy may place a patient at risk for sinusoidal obstruction syndrome, formerly known as veno-occlusive disease, which can present in a similar manner as Budd-Chiari syndrome (Table 159-4).

PHYSICAL EXAMINATION

With ALF, the most important step in the physical examination is assessing mental status and performing frequent neurological examinations. Clinicians must closely examine for any signs of chronic liver disease including spider angiomata, telangiectasias, Terry nails (fingernails and/or toenail appear white with a characteristic “ground-glass” appearance, with no lunula), palmar erythema, asterixis, splenomegaly, caput medusa, collateral venous patterns, and gynecomastia. Whereas ascites and varices are classically seen with chronic liver disease, these signs of portal hypertension can be seen with acute liver disease as well (eg, alcoholic hepatitis, venous thrombosis, presinusoidal portal hypertension).

In patients suspected of Wilson disease, a slit-lamp examination of the eye should be performed to assess for the presence of Kayser-Fleischer rings. In patients with intracranial hypertension from cerebral edema, Cushing triad—hypertension, bradycardia, irregular respirations—may or may not be present. Pupillary dilation or signs of decerebrate posturing typically occur late in the course of ALF and indicate a grave prognosis.

LABORATORY

A standard protocol of laboratory testing for patients admitted with ALF may facilitate a thorough evaluation, since the initial lab survey is extensive. Even if a diagnosis seems clear, evaluating for concomitant diseases such as viral hepatitis or human immunodeficiency virus is important in a patient who may require a liver transplant. Frequent hepatic function testing often necessitates indwelling vascular access (central venous catheter or arterial catheter) for serial blood draws and other clinical management (Table 159-5). No consensus guidelines recommend specific laboratory testing in the setting of probable ALF; however, laboratory testing is necessary in anticipation of possible liver transplantation.

Upon presentation, very high transaminase levels (AST and ALT levels >1000 U/L) may limit the differential to a few possible etiologies: acute viral infection, acetaminophen-induced injury, shock liver, or autoimmune hepatitis. AST and ALT levels exceeding 3500 U/L are highly correlated with acetaminophen poisoning. Whereas transaminase levels may be helpful initially, they do not add much information once the diagnosis of ALF is established. Transaminase levels that appear to be normalizing may indicate an obliterated liver, especially if associated with hypoglycemia and an unabated rise in the international normalized ratio. Alternatively, normalizing liver enzymes may also indicate improving hepatic injury facilitated by removal of a toxic exposure, natural history of a virus, or a treatment response (eg, autoimmune hepatitis, herpes simplex, hepatitis B). A bilirubin level on presentation carries prognostic significance, especially with drug-induced liver injury (Hy’s law, a prognostic indicator that suggests that drug-induced liver injury leading to jaundice, without hepatic transplantation, has a case fatality rate of 10%-50%). Initial INR and creatinine also provide prognostic information.

IMAGING

Ordering a multitude of imaging tests that require patient transport and movement may present unnecessary risk for a critically ill patient with ALF. Nevertheless, imaging—when safe to perform—may provide diagnostic information to direct management. All patients with ALF should undergo an abdominal ultrasound with Doppler to assess for signs of chronic liver disease and to assess the hepatic vasculature (hepatic veins, portal vein, hepatic artery). Computed tomography (CT) scan of the abdomen without contrast is preferred if the patient’s body habitus (eg, obesity, ascites, bowel distension) precludes an adequate ultrasound examination. Avoidance of IV contrast is important since many patients with ALF are at risk for developing acute kidney injury. Magnetic resonance imaging (MRI) of the abdomen with IV gadolinium is an alternative imaging modality that is center specific and depends on radiologic expertise.

In a patient with worsening mental status, a noncontrast brain CT will help to evaluate for intracranial hemorrhage or cerebral edema. Whereas cerebral edema, while prognostically ominous if seen on CT, may indicate the need for placement of an intracranial pressure monitor, recent studies suggest that this does not confer a significant mortality benefit, and in certain cases (nonacetaminophen ALF), it may be associated with worse outcomes. In a patient with suspected brain damage, transcranial Doppler ultrasounds or continuous electroencephalograms (EEGs) may provide prognostic information. To assess heart function and valves, a transthoracic echocardiogram may be necessary, especially if a patient is a possible liver transplant candidate. Plain films, such as chest x-rays and abdominal radiographs, may also be ordered as clinically indicated but are generally performed at the patient bedside (portable machine) to avoid unnecessary patient transport.

LIVER BIOPSY

When specific diseases such as autoimmune hepatitis, Wilson disease, HSV hepatitis, or lymphoma are suspected, a liver biopsy may be useful in assisting with diagnosis. A liver biopsy may also help determine the extent of necrosis and regeneration in cases of subacute liver failure. For the majority of ALF cases, a liver biopsy is not necessary and may even lead to complications such as hemorrhage or infection. In cases of severe ALF, a biopsy will most likely show extensive necrosis and “ghosts” of hepatocytes.

TRIAGE AND HOSPITAL ADMISSION

Any patient with suspected ALF should be admitted to the hospital for a thorough evaluation and consideration for liver transplant. If a patient is being admitted to a hospital that is not a transplant center, early transfer to a transplant center should be considered, especially if both encephalopathy and coagulopathy are present. Rapid deterioration in clinical and biochemical status following an initial stable presentation can occur. Some centers may use prognostic models to determine who should be transferred. At the present time, there are no validated criteria that dictate when to transfer a patient. Prior to transfer, a detailed communication between the receiving and transferring facility should address the existing neurologic, cardiopulmonary, hepatic, hematologic, and renal status. Prior to transfer, it may be necessary to electively intubate a patient for airway protection, as advancing encephalopathy can occur unexpectedly during transit.

Patients that develop grade II, III, or IV encephalopathy should be managed in an ICU (Table 159-6). Patients with grade III or IV encephalopathy may require both intubation for airway protection and possible ICP monitoring. Even though a patient may appear stable, rapid changes in clinical status may occur and should be dealt with in an urgent manner. Frequent neurologic evaluation, serial laboratory testing, and 24-hour monitoring in an ICU together provide the best opportunity to detect clinically significant changes in status.

MANAGEMENT

INITIAL MANAGEMENT (OVERVIEW)

A directed history and physical examination, extensive laboratory panel, and relevant imaging should be performed as quickly and safely as possible. A comprehensive history and physical examination needs to focus on the potential causes of liver injury, and any evidence of chronic liver disease that would preclude a diagnosis of ALF. Any barriers to potential liver transplant (comorbidities, lack of social support, active neoplastic or infectious processes) should be assessed.

In known or suspected cases of acetaminophen toxicity, acetaminophen nomograms are unreliable if the time from ingestion is not known, or if a single large ingestion was not taken. Activated charcoal may be given to patients within the first 4 hours of intentional acetaminophen overdose. Serial measurements of liver enzymes, coagulation parameters (INR and factor V levels), arterial pH, and arterial lactate will help to prognosticate the extent of liver damage. Serial assessments of mental status help determine stage of encephalopathy and direct further intervention (eg, head CT, ICP monitor) (Table 159-7).

ROLE OF N-ACETYL-CYSTEINE (NAC)

The administration of N-acetyl-cysteine (NAC) can provide protection from hepatotoxicity, especially if given within 12 hours of nonstaggered ingestion of acetaminophen. The main effects of NAC are to increase hepatic glutathione production and act as an antioxidant. The first dose of NAC may still be beneficial up to 48 to 72 hours following acetaminophen ingestion.

N-acetyl-cysteine may also offer benefit in non-acetaminophen causes of liver failure, as it seems to improve overall hemodynamic status and oxygen delivery to peripheral tissues. A recent trial randomized patients with non-acetaminophen ALF to intravenous NAC or placebo. Transplant-free survival improved in patients who received NAC. N-acetyl-cysteine also improved 1-year overall survival in the subset of patients with mild (grades 1 and 2) encephalopathy.

Based on available evidence, all patients with ALF, regardless of etiology, should receive NAC. In the absence of significant upper GI symptoms, NAC may be given orally or by NG tube as follows: 140 mg/kg (5% dextrose solution) loading dose followed by 70 mg/kg every 4 hours for 17 doses. Oral NAC has several potential side effects that include nausea, vomiting, rare urticaria, and bronchospasm. N-acetyl-cysteine may also be administered intravenously as follows: 150 mg/kg (5% dextrose solution) loading dose over 15 minutes, followed by 50 mg/kg given over 4 hours, followed by 100 mg/kg given over 16 hours. Intravenous NAC may lead to rare anaphylaxis, ECG abnormalities, bronchospasm, and false improvements in INR. The US Food and Drug Administration has not yet approved the intravenous use of NAC for ALF. N-acetyl-cysteine therapy can be stopped when the patient proceeds to liver transplantation, or when there is evidence that hepatic dysfunction is resolving, defined by improved encephalopathy and coagulopathy (INR < 1.5).

CONSULTATION

Depending on the severity and extent of multiorgan involvement, patients with ALF often require multiple consultations (Table ­159-7). A multidisciplinary approach is also very important for patients who may be candidates for liver transplantation. Consultations may include providers from critical care medicine, transplant hepatology (surgical and medical), anesthesiology, cardiology, nephrology, infectious diseases, pulmonary, psychiatry, social services, and possibly the ethics committee.

COMPLICATIONS AND PROGNOSIS

Since patients with advanced ALF present with multiorgan system failure, management should follow a system-based, goal-directed approach. Management of ALF in an organ-system-based manner requires multiple disciplines and coordinated care (Table 159-8).

CEREBRAL EDEMA AND INTRACRANIAL HYPERTENSION

Intracranial hypertension and cerebral edema remain serious complications of ALF, often leading to fatal herniation. Intracranial hypertension develops in approximately 20% to 30% of patients with ALF. Cerebral edema occurs more often in patients with a short time interval (28 days or less) between the development of jaundice and the onset of encephalopathy. Younger patients, with less cerebral atrophy than older patients, may not tolerate rapid shifts in ICP. Young age and high arterial ammonia correlate with increased risk of death from herniation. Cerebral edema is rarely seen in patients with mild (grades I and II) encephalopathy. Risk factors for progression to encephalopathy and cerebral edema include the presence of systemic inflammatory response syndrome, the need for vasopressors, the need for continuous renal replacement therapy (CRRT), or the presence of infection.

The exact mechanisms underlying elevated ICP in patients with ALF are not fully understood and are likely multifactorial. Both retrospective and prospective studies have shown that higher arterial ammonia levels portend a higher risk of cerebral herniation and death. An arterial ammonia >124 micromoles/L predicts mortality with a 77.5% diagnostic accuracy (sensitivity 78.6%, specificity 76.3%). Elevated ammonia levels lead to osmotic shifts that negatively affect astrocytes. When ammonia levels rise in subacute or chronic liver disease, there is time for adaptation. In ALF, the rapidity of the rise in serum ammonia is thought to limit the time for adaptation.

ICP MONITORING

Intracranial pressure monitoring is controversial. It allows measurement of cerebral perfusion pressure (CPP) and detection of ICH that is otherwise clinically silent. This may facilitate timely and rational use of ICP-lowering therapies, and allow optimization of ICP before and during liver transplantation. However, insertion of ICP monitors risks intracranial hemorrhage, and no mortality benefit has been demonstrated. A recent retrospective cohort study found that ICH was common in patients with ICP monitors, and hemorrhagic complications were rare (7%), but there was no 21-day mortality benefit in acetaminophen-related ALF and a worse prognosis in the nonacetaminophen group.

The utility of ICP monitors is extrapolated from the neurointensive care literature regarding management of traumatic brain injury. Clinical signs of elevated ICP are not always present, and pupillary dilation, decerebrate posturing, or changes on brain CT scans are often discovered late in the course. Therefore, a more accurate and potentially objective measure of ICP using a monitoring device may enable determination of ICH and guide decisions of therapy, prognosis, and candidacy for liver transplantation.

Significant ICH is defined as a sustained (>10 minutes) elevation of ICP >25 mm Hg. In addition to the ICP, the cerebral perfusion pressure—defined as the difference between the mean arterial pressure (MAP) and ICP (ie, CPP = MAP – ICP) with optimal CPP 60 to 80 mm Hg—guides patient management. Augmentation of MAP with vasoactive agents in order to maintain adequate CPP in the setting of an elevated ICP may be necessary.

Refractory ICH or decreased CPP are considered contraindications to liver transplantation in many centers. One study showed that prolonged high ICP (>25 mm Hg) and low CPP (<40 mm Hg) for greater than 2 hours portended poor neurologic recovery and should preclude liver transplantation. However, absolute pressure thresholds predicting irreversible neurologic damage or precluding transplantation should be used with caution, since such pressure thresholds have been challenged by reports of complete neurologic recovery after prolonged (>24 hours) ICH (ICP > 35 mm Hg) with low CPP (<50 mm Hg). For patients with ALF who are listed for transplantation, ICP monitoring may help with ICU management of cerebral edema and ICH prior to transplantation. Furthermore, since elevated ICP can persist perioperatively, ICP monitors can also be useful both during and after the liver transplant surgery. Given the absence of large randomized controlled trials, the use of ICP monitors currently remains center dependent.

The major risks involved with placing an intracranial device include infection and bleeding, with mortality from bleeding reported to be 1% to 3%. Intraventricular catheters bear greater risk than intraparenchymal or subdural catheters, which in turn bear greater risk than epidural devices. The intraparenchymal location appears to offer the optimal balance between accuracy and safety. Aggressive correction of coagulation parameters, along with the use of recombinant activated factor VII, may reduce bleeding risk and allow wider use of ICP monitoring devices.

MEDICAL MANAGEMENT OF ELEVATED ICP

Patients with ALF can progress rapidly from grade 1 to 4 encephalopathy. Frequent and detailed neurological examinations are vital, as subtle changes of increasing ICP can be missed. Once a head CT has ruled out any other process, patients with suspected ICH should have as little stimulation as possible. Chest physiotherapy and endotracheal suctioning may need to be minimized. Visitors, lights, and noises should be reduced. The head of the bed should be elevated to 30°. Cerebral perfusion pressure should be maintained at 60 to 80 mm Hg, and this may require cautious titration of MAP to compensate for increasing ICP. Trendelenburg position, head flexion, head rotation, and sudden change of position to supine should be avoided except when necessary for placement of a central venous catheter. Clinicians should avoid the placement of bilateral internal jugular lines for access, as this may collectively compromise intracranial venous drainage.

Lactulose may be used with early encephalopathy, but it lacks long-term benefit. Lactulose may also lead to gaseous distension of the bowel, which can complicate subsequent transplant surgery. The role of nonabsorbable antibiotics, such as rifaximin, has not been demonstrated.

Patients with grade III or IV encephalopathy should be intubated for airway protection. Endotracheal or intravenous lidocaine prior to intubation may be helpful in reducing spikes of ICP. For intubation, small doses of propofol may be used with caution, as the half-life of propofol is prolonged in hepatic failure. Propofol may also cause a decrease in MAP, thus compromising CPP.

Nonconvulsive seizure activity has been documented in a high proportion of patients with ALF and advanced stages of hepatic encephalopathy (HE). Seizure activity should be treated with ­phenytoin or fosphenytoin; however, there are insufficient data to recommend prophylactic anticonvulsants in all patients with ALF. It should be noted that propofol or benzodiazepine infusions used for sedation also provide potent antiseizure prophylaxis.

With ALF, the ICP should be maintained below 25 mm Hg in order to improve CPP. Treatment of elevated ICP is usually instituted when there is either a sustained rise in ICP (>10 minutes) or if clinical signs suggest ischemia or herniation. Increased sedation with propofol may be used for transient spikes of ICP due to agitation.

In the past, patients were mechanically hyperventilated in order to reduce the serum partial pressure of carbon dioxide (pCO₂) and cause cerebral vasoconstriction, thus reducing both cerebral blood flow and ICP. This effect is very transient, perhaps only lasting several minutes. Out of concern for cerebral hypoxia with reduced cerebral blood flow, hyperventilation is no longer a mainstay of therapy. Short-term hyperventilation has not been shown to improve outcome in ALF, but it does prolong survival in the ICU; therefore, it may occasionally be used as bridge therapy in selected patients who await a liver transplant.

The choice of pharmacologic agent for treatment of ICH is not standardized across transplant centers. Osmotic diuresis with IV mannitol effectively reduces cerebral edema and improves survival, based on a prospective, controlled study performed at King’s College in London. A bolus dose of 0.5 to 1 g/kg may be repeated once or twice as needed, as long as the serum osmolarity does not exceed 320 mOsm/L. Since mannitol is cleared by the kidneys, there is risk of intravascular hypervolemia and high-pressure pulmonary edema in the setting of renal failure. As mannitol is removed by dialysis, it may be used in patients on CRRT. Serum osmolarity and sodium level should be monitored every 6 to 8 hours when mannitol is used. In the absence of ICH, prophylactic mannitol is not indicated.

In recent years, hypertonic saline solution has become the more popular osmotic agent for hyperosmolar therapy. This growing popularity has come about in response to the complications associated with the use of mannitol, in particular acute renal failure and ICP rebound. Although an initial dose of mannitol may still be useful for acute control of elevated ICP, many centers move quickly to hypertonic saline as the mainstay of therapy. Induction and maintenance of hypernatremia may be used to prevent rising ICP. The efficacy of hypertonic saline in the prevention and treatment of ICH has been demonstrated in the neurologic intensive care literature. The goal of therapeutic hypernatremia is serum sodium of 145 to 155 mmol/L, which can be obtained by using 30% hypertonic saline in boluses of 20 mL. In contrast to mannitol, boluses of hypertonic saline have less tachyphylaxis to multiple doses. In the presence of CRRT, achieving the desired level of therapeutic hypernatremia may be a challenge, since a relatively hyponatremic replacement fluid is often used in the CRRT circuit.

Therapeutic cooling might protect against cerebral edema by various mechanisms: normalization of cerebral blood flow, decreased cerebral metabolic rate, decreased cytokine production, and decreased conversion of glutamate to glutamine, the compound that may precipitate cerebral edema in astrocytes. Decreased intracranial pressure and cerebral blood flow and increased cerebral perfusion pressure have been demonstrated with moderate (­32-35°C) hypothermia. Patients may be iatrogenically hypothermic from transfusions or a CRRT circuit. Cooling blankets can be added to further reduce core temperature. Hypothermia can also prevent ICP spikes during liver transplant surgery. Despite the potential utility of therapeutic hypothermia (TH), there is the concern that lowering the core temperature might lead to increased bleeding events, microbial infections, cardiac dysrhythmias, and potentially reduced hepatic regeneration. Currently, there are no controlled data on the effects of TH in ALF and no evidence-based guidelines for its use in these patients. A recent retrospective cohort study of ALF patients in the US Acute Liver Failure Study Group concluded that TH was not significantly associated with a 21-day spontaneous survival in non-APAP (acetaminophen) ALF patients. In APAP patients, TH may potential be associated with benefits in patients younger than 26 years of age. Additionally, the study concluded TH was not associated with increased bleeding or infection. The avoidance of hyperthermia is clearly a goal in management of ALF patients; however, a prospective trial is required to clarify the utility of TH in these patients.

In cases of elevated ICP refractory to osmotherapy (ie, mannitol and hypernatremia) and hypothermia, induction of a barbiturate coma using thiopental or pentobarbital can be considered. Significant myocardial depression and systemic hypotension may limit the use of a barbiturate coma. Some ICUs have incorporated the use of neuromuscular paralysis to optimize patient-ventilator synchrony to prevent any ICP surges due to patient-ventilator dyssynchrony.

A continuous electroencephalogram can monitor burst suppression, the primary goal of either coma induction or paralysis with sedation.

Finally, intravenous indomethacin induces cerebral vasoconstriction and may temporarily reduce ICP; however, this medication is not easily prepared and it may be quite expensive for only a transient benefit.

AIRWAY AND VENTILATION

Patients with ALF who reach grades III to IV encephalopathy require intubation and mechanical ventilation. When intubating, a clinician skilled in the procedure should maintain adequate preoxygenation, prevention of hypercapnea, and avoidance of hypotension. Propofol sedation and lidocaine (endotracheal or intravenous) may help to minimize spikes in ICP. Pre-intubation, clinicians may utilize neuromuscular blockade. When indicated, cis-atracurium—a nondepolarizing agent, metabolized independently of liver or renal function—is the paralytic of choice and permits neurologic assessment 40 to 60 minutes after the bolus. No consensus drives practice regarding the specific mode of mechanical ventilation; therefore, either volume or pressure controlled mode can be utilized. Initial ventilator settings should include a low tidal volume strategy of 6 mL/kg of ideal body weight and a maximum plateau pressure of 30 cm H₂O, in order to minimize the effects of mechanical ventilation-induced barotrauma. If a low tidal volume strategy increases pCO₂, this may lead to cerebrovascular dilation and increase in ICP. Therefore, clinicians should strictly manage pCO₂ through minute ventilation and adjust respiratory rate accordingly.

Optimal patient-ventilator synchrony with adequate sedation and analgesia helps to prevent spikes in ICP. Critical care physicians should employ paralysis if patient-ventilator asynchrony occurs despite aggressive sedation and analgesia. While mechanical hyperventilation does not improve outcomes, spontaneous hyperventilation should not be inhibited, as it does result in mild hypocapnia and cerebrovascular constriction, thus reducing ICP. Whenever possible, to avoid any spikes in ICP, bronchoscopy should be avoided in patients with ALF.

CIRCULATION

Circulatory changes associated with ALF can resemble changes seen with endotoxinemia-induced septic shock. The pattern observed depends on the etiology and rapidity of ALF. In hyperacute liver failure due to acetaminophen overdose, patients can develop fulminant peripheral cardiovascular collapse that can lead to early death. Early hemodynamic changes in ALF include increased portal pressures, splanchnic sequestration of blood, and decreased central venous return. Fluid resuscitation and vasopressor support may be necessary, but fluid requirements in ALF may be less than in patients with septic shock. Some experts recommend placement of a pulmonary artery catheter to aid in assessing volume status. NAC benefits hemodynamic stabilization and improves oxygen consumption in patients with ALF.

In order to maintain an adequate CPP, systemic vasopressor support may be necessary, especially if the patient is hypotensive from ALF-induced vasodilation. Clinicians should maintain close monitoring and titration of oxygen delivery while using vasoconstrictors. A mixed, or central, venous oxygen saturation can guide the institution and titration of vasoactive agents.

Careful pressor selection facilitates adequate blood flow to the liver and brain. Norepinephrine serves as the vasopressor of choice in cases of ALF. It increases hepatic blood flow in parallel with cardiac output, and has less tachycardia than dopamine. Norepinephrine also leads to less lactate production than epinephrine. Alpha-adrenergic agents (epinephrine, high-dose vasopressin, terlipressin) can worsen peripheral oxygen delivery and lead to splanchnic ischemia. Terlipressin increases both cerebral blood flow and ICP, due to breakdown of autoregulation.

Relative adrenal insufficiency can occur with ALF, and stress dose steroids (hydrocortisone up to 200-300 mg daily) have been shown to reduce norepinephrine requirements. Intravenous steroids should be given to patient with ALF who develop persistent hypotension despite adequate IV fluids and vasopressor support.

Hemodynamics and intracranial hypertension frequently improve after removal of the native liver during liver transplant surgery. Therefore, ALF patients who are listed for transplantation but who have persistent circulatory dysfunction despite maximal medical measures, should be considered for a bridging, last-resort hepatectomy.

RENAL FAILURE

Acute renal failure occurs frequently in the setting of ALF, with an incidence up to 70%, and portends a poor prognosis. Causes of renal failure may include a prerenal insult (systemic vasodilation, hypovolemia, sepsis), direct renal injury (acute tubular necrosis from hypoperfusion, acetaminophen, or other toxins), or postrenal obstruction. The nephrotoxic effects of acetaminophen usually reverse spontaneously, but whether NAC contributes has not been proven.

Based on randomized trials, early initiation of CRRT improves outcomes over intermittent hemodialysis in order to reduce dramatic fluid shifts and avoid spikes in ICP. High-volume hemofiltration improves laboratory (lactate, base deficit) abnormalities and reduces vasopressor requirements, but may not affect survival in ALF. Because patients with ALF do not tolerate a lactate load, bicarbonate buffered hemofiltration is recommended.

BLEEDING AND CLOTTING

Coagulation abnormalities, due to both decreased synthesis and increased consumption, occur frequently in patients with ALF. Thrombocytopenia occurs in 50% to 70% of patients, but only 5% of patients with ALF experience significant bleeding. Variceal bleeding almost never occurs, as chronic portal hypertension is absent. Spontaneous intracranial bleeding (in the absence of an ICP monitor) is rare as well. Practice-based recommendations for platelet transfusion in ALF include (1) all patients with platelet count < 10,000/mm³, (2) patients with sepsis and platelet count < 20,000/mm³, or (3) patients with a planned invasive procedure or active bleeding and a platelet count < 50,000/mm³. Patients with ALF also bear increased thrombosis risk due to reduced protein C and protein S production; therefore, venous thromboembolism prophylaxis with subcutaneous heparin may be instituted.

Hepatically synthesized factors (factor V, factor VII) and INR provide the most sensitive biochemical indicators of hepatic synthetic function and recovery. Therefore, administration of fresh frozen plasma (FFP)—which can obscure the picture of endogenous coagulation factor production—should be avoided in the absence of active bleeding. Vitamin K (5-10 mg subcutaneously daily for 3 days) can help in eliminating any component of malnutrition as a source for coagulopathy. Fresh frozen plasma transfusion and plasmapheresis carry additional risks of volume overload and transfusion-related acute lung injury. If an invasive procedure, such as an ICP monitor, is planned, activated recombinant factor VII can temporarily correct the INR. A minimum dose of 40 μg/kg is recommended in order to achieve a transient (therapeutic window of 60-90 minutes) reversal of coagulopathy. Repeat dosing can be administered as necessary. Factor VIIa has been associated with rare cases of thromboembolism and cerebrovascular accident. Bleeding rarely poses a major problem with transplant surgery itself, and the preoperative degree of coagulopathy does not predict post-transplant outcome.

INFECTION

Patients with ALF have altered immune function and, in the setting of multiple intravenous lines, catheters, and devices, have ­moderate-to-high risk for bacterial and fungal sepsis. Up to 80% of ALF patients develop bacterial infections, and 30% of patients develop fungal infections. Most infections are diagnosed within 72 hours of hospital admission and are commonly localized to the respiratory tract, urinary tract, or bloodstream. Gram-positive infections with Staphylococcus or Streptococcus species occur more commonly than Gram-negative or Candida infections.

Clinicians should maintain a low threshold to evaluate for infection since the usual signs of infection may be absent in nearly one-third of individuals with ALF. Because active infection may preclude liver transplantation, some centers advocate starting empiric antimicrobial therapy upon hospital admission. While antibiotic prophylaxis reduces the incidence of infection and may reduce the risk of ICH and cerebral edema, such prophylaxis has not shown an overall survival benefit. Empiric antibacterial and antifungal therapy should be administered in the setting of shock, and may be considered in advanced encephalopathy or in patients listed for transplantation.

METABOLIC ABNORMALITIES AND NUTRITION

Metabolic derangements are common in patients with ALF. Alkalosis and acidosis may both occur and are best managed by identifying and treating the underlying disorder. Hypoglycemia should be managed with continuous glucose infusions and frequent serum glucose checks. Sodium, potassium, and magnesium levels are often low and should be supplemented. Hypophosphatemia is common and, while necessitating supplementation, may be a good prognostic sign, as it may reflect the consumption of ATP in a regenerating liver. High or normal phosphate levels may indicate a lack of hepatic regeneration and renal impairment. ALF patients frequently have increased energy expenditure and a catabolic state. Therefore, in order to reduce infectious complications, early enteral feedings should be initiated, including 60 g of protein per day. Enteral feedings may also reduce the risk of GI bleeding due to stress ulcerations in critically ill patients.

LIVER SUPPORT SYSTEMS AND AUXILIARY GRAFTS

The MARS^® (Molecular Adsorbent Recirculating System) device is approved by the Food and Drug Administration for use in the US in the treatment of acute liver failure from drug overdose and poisonings as well as HE due to a decompensation of a chronic liver disease. MARS combines traditional continuous renal replacement therapy technology with large protein-bound particle removal via albumin dialysis. It removes ammonia, bilirubin, bile acids, aromatic amino acids, nitric oxide, tryptophan, copper, creatinine, protoporphyrin, urea, and diazepam. Human albumin cleans protein-bound toxins, while the bicarbonate-based dialysate binds other water-soluble elements. This reduces plasma toxicity, improves patient clinical conditions (hemodynamics, hepatic encephalopathy, urine output), enhances the regeneration of liver cells and may help to recover native liver functions.

For patients in whom a liver graft donor has been identified, total hepatectomy may be considered, as it removes the major source of proinflammatory cytokines that contribute to cerebral vasodilation. Auxiliary liver transplants refer to reduced size graft from living or deceased donor. They have been used in Europe to provide temporary hepatic function as a bridge to recover of the native liver. The best outcome has been seen in patients younger than 40 years of age with either acute viral hepatitis or acetaminophen toxicity.

PROGNOSIS

Regardless of the acuity, the etiology of ALF best predicts the likelihood of spontaneous recovery without transplantation. Patients with ALF due to acetaminophen, hepatitis A, shock liver, or pregnancy-related disease have > 50% transplant-free survival, whereas patients with other etiologies have a < 25% transplant-free survival. Patients with autoimmune hepatitis, acute hepatitis B, or acute HSV may improve with specific therapy and not require a liver transplant.

Clinicians should use accepted prognostic models to evaluate patients with ALF. The King’s College prognostic criteria separate patients with acetaminophen-induced ALF from those with non–acetaminophen-induced ALF (Table 159-9).

The King’s College model has high specificity (92%) in determining who will need a liver transplant, but low sensitivity (69%). If a patient fulfills the King’s College criteria, the likelihood ratio positive for a poor prognosis without transplant is 8.63. On the other hand, the likelihood ratio negative for a patient with unfulfilled King’s College criteria is 0.34. Therefore, the King’s College criteria may misclassify a substantial number of patients who will die without a liver transplant. A recent meta-analysis found that the King’s College criteria and a pH < 7.30 were both fairly specific in predicting a poor outcome.

One small study showed that an APACHE II score > 15 on admission for patients with acetaminophen-induced ALF shows a specificity of 92% and a sensitivity of 81% (likelihood ratio positive 10.1, likelihood ratio negative 0.21) for predicting need of liver transplant. The Model for End Stage Liver Disease (MELD) score may eventually prove to be superior to the King’s College criteria in cases of non–acetaminophen-induced ALF. A single study in Argentina showed a MELD score of greater than 30 to be superior to the King’s College criteria in predicting poor prognosis without transplant in patients with non–acetaminophen-induced ALF. The positive predictive value and negative predictive value of a MELD > 30 was found to be 91% and 100%, respectively (diagnostic accuracy 95%). Further studies are needed for validation.

Finally, multiple clinical predictors correlate with poor outcome in the absence of transplant. Encephalopathy at the time of admission predicts a worse prognosis. Patients presenting with grade III or IV encephalopathy die or require transplant in 60% to 90% of cases. Age, renal failure, and severity of coagulopathy are also independent predictors of mortality without transplant. Even the King’s College criteria lack the sensitivity to replace an experienced transplant physician’s clinical judgment.

DISCHARGE PLANNING

If a patient recovers from ALF with or without transplant, he or she should receive education about the disease and ways to prevent future injury. Depending on the etiology of ALF, ongoing treatment and follow-up may be necessary. For example, a patient with autoimmune hepatitis will need close hepatology follow-up and treatment with immunosuppressant medications. Also, a patient with acute-on-chronic hepatitis B will require hepatology follow-up, not only for antiviral treatment but also for hepatocellular carcinoma screening. The need for rehabilitation depends on overall performance and details of the hospitalization.

If a patient undergoes a liver transplant, a strong social network is vital, and close follow-up in the transplant center must be arranged. Rehabilitation or a skilled nursing facility may be necessary for some patients, especially if social support becomes a limitation. In contrast to patients transplanted for chronic liver disease, social support breakdown occurs more commonly when a patient undergoes liver transplantation for ALF, likely due to less time for education and a comprehensive assessment.

QUALITY IMPROVEMENT

Since fatal outcome occurs frequently in ALF without liver transplant, and because liver transplant is not always an option (eg, organ availability, poor candidate), public health measures should focus on prevention of ALF. Controlling distribution of acetaminophen could reduce overdoses. Some countries limit acetaminophen access by regulating the number of blister packs that can be purchased. Education about unintentional acetaminophen overdose, and education about drug and alcohol interactions, may help to reduce acetaminophen-related ALF in the United States. Finally, early recognition of drug-induced hepatoxicity, along with cessation of offending agents, may prevent some cases from progressing to liver failure.

In some cases, patients with ALF who are good candidates for liver transplant die before a donor liver becomes available. Increasing organ donation will lead to more lives saved, and additional education and outreach programs to increase liver donation are needed. Also, other creative ways to increase the donor pool (eg, extended criteria donors, paired exchanges, and partial-liver living donors) may augment the pool of available transplant organs.

Due to the low incidence of ALF, there are few controlled trials that can help to guide management (Table 159-10). Standards of intensive care in this patient population have not been well established and are mostly guided by expert opinion. Management may differ from center to center. Multicenter collaborations and prospective trials may improve future management of patients with ALF.

SUGGESTED READINGS

INTRODUCTION

Cirrhosis of the liver is a chronic illness that progresses at a variable rate, dependent on the etiology and the activity of the offending toxin. It is a dynamic process, which is potentially reversible in the earlier stages if the offending agent is either removed or modified. Classically, cirrhosis has been defined histologically as architecturally abnormal nodules separated by bands of fibrous tissue. The diagnosis has usually been based on clinical suspicion, confirmed by radiologic studies and a liver biopsy as the gold standard. ­However, sampling error on biopsy may lead to incorrect staging of the disease, where the transition from severe fibrosis to cirrhosis is misinterpreted. Therefore, advanced chronic liver disease has been proposed as a more encompassing definition for the transitional stages. However, in this chapter, we will continue to use cirrhosis as the term to characterize this entity.

Patients with cirrhosis may be further classified as compensated or decompensated. Decompensation is defined as the appearance of ascites, variceal bleeding, hepatic encephalopathy or jaundice. A prognostic classification consisting of five stages has been proposed to account for the increasing mortality with progressive decompensation (Figure 160-1). Stage one includes patients with compensated cirrhosis that do not have any complications such as ascites or esophageal varices. In stage two, patients have developed nonbleeding esophageal varices but are still considered compensated. Stage three describes patients who have transitioned to decompensation, characterized by the development of hemorrhage from esophageal varices. Patients may further progress to stage four which is characterized by the development of nonbleeding complications, including the development of ascites, hepatic encephalopathy or the hepatorenal syndrome. Stage five includes patients who have developed a second decompensating event. In a recent study by D’Amico et al that included 494 patients with alcoholic, viral or cryptogenic cirrhosis, the initial decompensating event was ascites (33%), variceal bleeding (10%), and hepatocellular carcinoma (9%), while encephalopathy or jaundice as the first decompensating event was rare. In this study, the 5-year mortality was 1.5% for patients in stage one compared to 88% for patients in stage five. In addition to describing the natural history of the disease process, a potential benefit of this staging system is the ability to target specific therapy based on the prognosis at a given stage. For example, patients who presented with variceal bleeding without ascites had a significantly better survival than those in whom ascites was present at the onset of the bleeding event and would be less likely to be considered as candidates for liver transplantation.

PATHOGENESIS

The major complications of cirrhosis are secondary to the development of portal hypertension. The initiating event is an increase in intrahepatic and portocollateral resistance, followed by an increase in portal venous flow. The increase in intrahepatic resistance has relatively fixed components including sinusoidal encroachment, collagen deposition, vascular tree pruning and nodular formation. Dynamic components of intrahepatic resistance are governed by an overexpression of endogenous vasoconstrictors, especially endothelin, and a diminished expression of vasodilators, primarily nitric oxide. The increase in portal blood flow is related to systemic vasodilation secondary to increased systemic production of vasodilators, primarily nitric oxide but also including increased production of glucagon, tumor necrosis factor, prostaglandins and other cytokines. The resulting hyperdynamic circulation leads to an increase in portal venous blood flow against intrahepatic resistance, resulting in portal hypertension. Portal hypertension is further aggravated by an increase in angiogenesis which exacerbates the increase in portal pressure and promotes the development of a portosystemic collateral circulation including the development of esophageal varices.

CLINICAL PRESENTATION

Presenting symptoms of cirrhosis are nonspecific and include the development of fatigue and lethargy. Physical findings include jaundice, peripheral muscle wasting, splenomegaly, spider angiomata, abdominal collateral vessels and an altered mental status. However, patients presenting initially with cirrhosis often have none of these findings.

Laboratory findings include an increase in serum bilirubin, a decrease in serum albumin, increases in AST, ALT and alkaline phosphatase and prolongation of the prothrombin time. A decrease in platelet count in the presence of cirrhosis suggests the development of portal hypertension but the variability of this finding makes it unreliable as a screening test.

DIAGNOSIS OF CIRRHOSIS

Although liver biopsy remains the gold standard for the diagnosis of cirrhosis, the test is invasive, often limited by inadequate sample size and prone to sampling error. Recently, there is increased interest in noninvasive means for diagnosing cirrhosis. Most prominent of the noninvasive methods is the use of transient elastography (TE) (fibroscan) as a means of identifying patients with cirrhosis who are at risk of developing clinically significant portal hypertension. The technique is easy to perform at bedside, has excellent reproducibility, and has very good correlation with invasive measurements of portal pressure (eg, the hepatic venous pressure gradient [HVPG]), especially in patients in the compensated stages of cirrhosis. ­Limitations include occasional false positives and unreliability in obese patients, patients with clinically evident ascites and patients with acute liver failure. The fibroscan has recently received Food and Drug Administration (FDA) approval and has become a standard diagnostic test in liver centers throughout Europe and North America.

Other noninvasive methods for the diagnosis of cirrhosis include serum biomarkers (eg, Fibrotest) which have gained popularity in Europe, more so than in the United States. A number of radiologic approaches are under investigation including MR elastography, ultrasound elastography and measurements of splenic stiffness but these should still be considered experimental techniques.

Portal pressure as measured by the HVPG provides excellent information for diagnosis, prognosis and management of portal hypertension. Hepatic venous pressure gradient measurements are excellent in predicting the development of esophagogastric varices, ascites, esophageal variceal hemorrhage, hepatic decompensation and the development of hepatocellular carcinoma. Hepatic venous pressure gradient measurements are especially useful in assessing response to pharmacologic therapy of portal hypertension. ­However, because of the invasive nature of these measurements, they are primarily used in clinical research studies and have limited usage in clinical practice.

ESOPHAGOGASTRIC VARICES

Patients with cirrhosis develop esophageal varices at a rate of 8% per year. The risk of hemorrhage from varices is 5% to 15% per year, with patients having large varices at greater risk. Variceal hemorrhage is associated with a 15% to 20% 6-week mortality. If the index bleed is controlled and no further treatment to prevent rebleeding is initiated, the risk of recurrent hemorrhage is 60% to 70% with most episodes occurring within 1 year of the index bleed.

SCREENING FOR ESOPHAGEAL VARICES

Current practice guidelines recommend screening by esophagogastroduodenoscopy (EGD) for the presence of varices in all patients with the diagnosis of cirrhosis. In patients with compensated cirrhosis and no varices on the initial screening EGD, the EGD should be repeated in 2 to 3 years unless there is evidence of clinical progression of the disease. In patients with decompensated cirrhosis and no varices on the initial examination, EGD should be repeated yearly. Once varices have been detected, further EGD is not indicated unless there is evidence of gastrointestinal bleeding. Because of cost and patient preference considerations, there is considerable interest in noninvasive techniques that might identify patients with cirrhosis at low risk for the development of varices. Noninvasive serum markers combining a number of biochemical tests may identify patients at low risk for developing varices but their positive and negative predictive values are insufficient to avoid screening endoscopy. A low platelet count (<100,000/mm²) has been suggested for screening patients at risk for varices but the predictive accuracy is insufficient to defer patients from endoscopic screening. The platelet count/spleen diameter ratio as measured by abdominal ultrasound is somewhat better at identifying patients at risk. The best noninvasive test for identifying patients at low risk for the development of varices is the measurement of liver stiffness by transient elastography. In patients with TE values <10 kPa in the absence of clinical signs of portal hypertension, the risk of developing esophageal varices is extremely low and screening endoscopy may be avoided. Until more data are available, patients with a TE value >10 kPa should be screened by endoscopy for varices. If the TE value is >15 kPa, patients are at high risk for developing varices. Wireless capsule endoscopy has 85% to 95% accuracy in determining the presence of varices, variceal size and endoscopic red color signs. However, this technique is not widely available and it has not replaced endoscopic screening as the standard diagnostic procedure.

RISK STRATIFICATION FOR VARICEAL HEMORRHAGE

With rare exceptions, patients with cirrhosis require an HVPG >10 mm Hg for the development of esophageal varices and an HVPG >12 mm Hg for variceal hemorrhage to occur. Risk factors for variceal hemorrhage as identified by endoscopy include the size of varices (large varices are at greater risk than small varices) and the endoscopic red color signs. The latter include red wale marks that appear as longitudinal red streaks on varices, hematocystic spots which look like small varices on the surface of larger varices and cherry red spots. In addition, the worse the Child-Pugh score, the greater the risk of bleeding. In patients with alcoholic cirrhosis, active alcohol consumption increases the risk of hemorrhage.

TREATMENT OF ESOPHAGEAL VARICES AND VARICEAL HEMORRHAGE

Options for the treatment of esophageal varices include pharmacologic agents that decrease portal pressure, endoscopic techniques that obliterate the varices and TIPS (transjugular intrahepatic portosystemic shunt) or surgical shunts that achieve a major reduction in portal pressure.

PHARMACOLOGIC THERAPY

The goal of pharmacologic therapy is to reduce portal venous blood flow and/or intrahepatic and portocollateral resistance in order to achieve a decrease in portal pressure. The primary drugs to achieve this goal on a long-term basis are the nonselective β-blockers (propranolol, nadolol, and timolol) and the long acting nitrate, isosorbide-5-mononitrate (ISMN).

β-Blockers act by decreasing cardiac output via blockade of the β1 receptors and by blockade of the β2 receptors, leaving unopposed alpha adrenergic activity which causes vasoconstriction of the splanchnic vessels. These drugs are given orally and should be titrated to the maximally tolerated dose. Unfortunately, only 35% to 40% of patients treated with β-blockers achieve a clinically significant hemodynamic response, defined as a 20% reduction in HVPG or a reduction in HVPG to below 12 mm Hg. Carvedilol, a nonselective β-blocker with intrinsic anti-alpha-1 adrenergic activity has been shown to produce a greater reduction in HVPG than the nonselective β-blockers when given as monotherapy.

Isosorbide-5-mononitrate increases the delivery of nitric oxide to the intrahepatic circulation, thus decreasing intrahepatic resistance. It also decreases cardiac output by decreasing cardiac preload and enhances systemic vasodilation and venous pooling which causes a reflex splanchnic arterial vasoconstriction. When given together with nonselective β-blockers, there is an incremental reduction in HVPG. Isosorbide-5-mononitrate is clinically ineffective when given as monotherapy.

Simvastatin has been shown to produce an incremental decrease in the HVPG when combined with nonselective β-blockers and prospective trials are currently underway to determine its clinical efficacy.

The major side effects of nonselective β-blockers include hypotension, fatigue, lethargy, depression and dyspnea in patients with pulmonary disease. Their use is contraindicated in patients with restrictive airway disease, congestive heart failure, bradycardia and heart block. Their use in insulin dependent diabetics needs to be monitored carefully. As a result, 15% to 20% of patients with cirrhosis have contraindications to the use of β-blockers, and an additional 15% develop severe side effects leading to discontinuation of the drug. The major side effects of ISMN are systemic arterial hypotension and headaches.

The drugs used in the management of acute variceal bleeding include the vasopressin analog, terlipressin and somatostatin and its analogs, octreotide, vapreotide and lanreotide. The goal of these drugs is to reduce portal venous blood flow and portal pressure Terlipressin achieves a greater reduction in portal pressure than the somatostatin analogs. The drugs are safe with minimal side effects (Table 160-1).

Unfortunately, none of these drugs (mostly generic) have received FDA approval specifically for the treatment of esophageal varices and variceal hemorrhage. The nonselective β-blockers and octreotide are used off label. Terlipressin is widely used in Europe but as yet, has not received FDA approval for use in the US.

ENDOSCOPIC VARICEAL LIGATION (EVL)

Endoscopic variceal ligation involves the placement of rubber bands on the variceal columns via an endoscope that causes mucosal and submucosal necrosis leading to the formation of scar tissue and the obliteration of the varix. It is a local treatment and does not significantly alter portal hypertension. The most serious side effects are possible perforation of the esophagus and the formation of mucosal ulcers that can lead to significant hemorrhage. Endoscopic variceal ligation has replaced sclerotherapy as the endoscopic treatment of choice.

TRANSJUGULAR INTRAHEPATIC PORTOSYSTEMIC SHUNT (TIPS)

The TIPS procedure involves placement of a catheter into the right internal jugular vein and advancing it via the hepatic vein through the liver to form a tract connecting to the portal vein. This transhepatic tract is then dilated and a flexible coated metallic stent is placed, resulting in a shunt between the hepatic and portal veins and a significant decrease in portal pressure. The procedure is performed by interventional radiologists and has replaced portosystemic shunt surgery in most institutions.

Short-term complications of the TIPS procedure include fever, infection, renal dysfunction, intrahepatic and intraperitoneal hemorrhage and progressive hepatic failure. The major long-term complication is the development or worsening of hepatic encephalopathy, which has an incidence of approximately 30% to 35%. Risk factors for the development or worsening of hepatic encephalopathy after TIPS include increasing age, the presence of minimal hepatic encephalopathy at the time of TIPS placement, and a history of overt hepatic encephalopathy. The severity of encephalopathy correlates with the diameter of the shunt. With the advent of coated stents, shunt occlusion is much less of a problem. Doppler sonography is a noninvasive method to monitor shunt patency and is usually performed at 6-month intervals for the first 2 years after TIPS placement.

PREVENTION OF FIRST VARICEAL HEMORRHAGE (PRIMARY PROPHYLAXIS)

Based on the results of a large, long-term, multicenter randomized controlled trial (RCT), there was no benefit for the use of nonselective β-blockers in the prevention of the development of esophageal varices or other complications of portal hypertension in patients with compensated cirrhosis and an elevated HVPG.

Because of the high risk of bleeding and the associated mortality, treatment to prevent initial variceal hemorrhage is mandatory. ­Current practice guidelines (AASLD, ACG, Baveno) recommend the use of either nonselective β-blockers or EVL for the prevention of variceal hemorrhage in patients with large varices or small varices in patients with Child-Pugh class C cirrhosis or in patients with small varices when red color signs are present on endoscopy. In other patients with small varices, β-blockers may be used, but more data are needed to establish their efficacy. A meta-analysis of 11 RCTs totaling 1190 patients showed a reduction in first variceal hemorrhage in patients with large varices from 30% in the control groups to 14% in the treatment groups. Randomized controlled trials showed similar results when comparing EVL to control groups. ­Neither treatment has been shown to improve survival. In well-designed RCTs with reasonable sample sizes, β-blockers and EVL were equally effective. As they are comparable in efficacy, the choice should be based on local expertise, side effects of the treatment and patient preference. In patients in whom varices have been eradicated with EVL, surveillance endoscopy should be performed because of the risk of recurrent variceal formation. However, EVL only reduces the risk of variceal bleeding while nonselective β-blockers reduce the risk of other complications of portal hypertension (ie, ascites) in addition to variceal bleeding.

TREATMENT OF ACUTE VARICEAL HEMORRHAGE

In patients with acute upper gastrointestinal bleeding, the source of bleeding should be determined by endoscopy. If varices are suspected in patients with cirrhosis, the patient should be started on a vasoactive drug (eg, terlipressin, octreotide). The vasoactive drugs should be continued for up to 5 days. Intubation to protect the airway is recommended in patients with active bleeding or altered mental status. In the absence of contraindications (eg, QT prolongation), infusion of erythromycin 250 mg IV 30 to 120 minutes prior to endoscopy should be considered (Figure 160-2).

A hemoglobin of 8.0 g/dL or a hematocrit of 25% should be the goal of transfusion in order to avoid an increase in portal pressure and exacerbation of the bleeding episode associated with overtransfusion. If the platelet count is greater than 50,000, platelet transfusions are not warranted. If the platelet count is lower, the use of platelet transfusions is controversial and their use has not been shown to affect the bleeding episode. The use of fresh frozen plasma (FFP) or recombinant factor VII transfusions is also controversial without clear evidence of their efficacy.

In patients with acute variceal bleeding, infection has been associated with early rebleeding and a high mortality rate. Based on the results of several RCTs, the short-term use of prophylactic antibiotics is mandatory. Preferred antibiotics include norfloxacin (400 mg twice daily for 7 days) or quinolone antibiotics (eg, levofloxacin, ciprofloxacin). In patients with decompensated cirrhosis, intravenous ceftriaxone (1.0 g/d) has been shown to be superior to norfloxacin. As there is increasing hospital based drug resistance to quinolones, this should be considered in the choice of antibiotics. If there is evidence of hepatic encephalopathy, institution of lactulose or rifaximin is recommended.

Once esophageal varices have been established as the source of bleeding, esophageal variceal ligation should be instituted to control the bleeding episode, EVL is successful in up to 90% of patients. If the bleeding remains uncontrolled or if there is severe early rebleeding, TIPS should be considered as the next intervention. Balloon tamponade with the Sengstaken-Blakemore tube or the placement of an esophageal stent can be used for temporary control of bleeding as a bridge to TIPS. Balloon tamponade should be limited to 24 hours as the complication rate increases significantly with longer use. Esophageal stents may be used for a longer period and are generally safer.

In patients with Child-Pugh class C cirrhosis or Child-Pugh class B with active bleeding, the placement of TIPS within the initial 24 hours from the onset of bleeding has been shown in a large multicenter RCT to be superior to medical therapy with EVL and vasoactive drugs and is associated with better control of the bleeding episode and improved survival.

PREVENTION OF RECURRENT VARICEAL HEMORRHAGE (SECONDARY PROPHYLAXIS)

As there is a 60% chance of recurrent hemorrhage after control of the acute bleeding episode, treatment to prevent recurrent hemorrhage is mandatory. Current practice guidelines recommend a combination of nonselective β-blockers and EVL as the treatment of choice. The addition of ISMN may offer a slightly lower rebleeding rate but has the risk of increased side effects. EVL should be performed every 1 to 2 weeks until varices have been obliterated. Endoscopy should be repeated every year to screen for recurrent varices. A meta-analysis of 23 RCTs comparing combination therapy to endoscopic monotherapy shows a reduction in recurrent variceal bleeding from 37.5% for endoscopic monotherapy to 25.3% for combination treatment. There is also a modest improvement in survival. If bleeding continues to recur, TIPS is the next option. Transjugular intrahepatic portosystemic shunt is very effective in preventing recurrent hemorrhage but is associated with an increased risk of hepatic encephalopathy. Surgically performed shunts are still an option in a few centers, but their use has declined significantly with the availability of TIPS. Variceal bleeding by itself is not an indication for liver transplantation.

GASTRIC VARICES

Gastric varices account for 10% to 15% of variceal bleeds and are associated with a higher rate of rebleeding. They occur more commonly in patients with extra-hepatic portal and/or splenic vein obstruction (eg, can be found in patients with chronic pancreatitis). Gastric varices can be divided into two types, those that are contiguous with esophageal varices (GOV1) and those that are located in the fundus of the stomach (GOV2, IGV1) or elsewhere in the stomach (IGV2). GOV1 varices are treated with EVL or pharmacologic therapy, similar to esophageal varices. The isolated gastric varices are best treated via endoscopic injection of a cyanoacrylate glue. Side effects of glue installation include fever, sepsis, retroperitoneal abscess formation and embolization. For failures of endoscopic treatment, control of hemorrhage by TIPS is successful in 90% of cases. Urgent TIPS should be considered in any patient with massive hemorrhage from gastric varices or early recurrent hemorrhage after failure of endoscopic therapy.

PORTAL HYPERTENSIVE GASTROPATHY

As seen on endoscopy, portal hypertensive gastropathy is characterized by erythema of the gastric mucosa in a mosaic like pattern with mucosal congestion and dilated capillaries and venules in the mucosa and submucosa of the stomach. It is present in more than half of patients with cirrhosis and portal hypertension and usually presents as a chronic anemia.

Hemorrhage from portal hypertensive gastropathy accounts for about 8% of gastrointestinal hemorrhage in patients with cirrhosis and is usually mild and self-limited. Nonselective β-blockers are the treatment of choice for prevention of recurrent hemorrhage. In patients with more massive bleeding requiring transfusions or in whom medical therapy has failed, TIPS should be considered.

PORTAL VEIN THROMBOSIS

Patients with portal vein thrombosis as the cause of portal hypertension should be evaluated for inherited and acquired thrombophilic factors, paroxysmal nocturnal hemoglobinuria, autoimmune disorders and myeloproliferative neoplasia. This is best managed in conjunction with a hematologist. For patients with acute primary thrombosis of the portal vein and/or the splanchnic venous circulation, anticoagulation is safe and recommended. Low molecular weight heparin and vitamin K antagonists are the treatment of choice. There is limited data on the use of long-term anticoagulation but it is probably safe.

ASCITES

Ascites is the pathologic accumulation of fluid within the peritoneal cavity. The most common cause of ascites is cirrhosis, which accounts for approximately 80% of cases. Other common causes include malignancy and congestive heart failure. Approximately 5% of patients with ascites have “mixed ascites” defined as ascites formation from two or more causes.

Ascites represents both the most common complication of portal hypertension and the leading cause for hospitalization in cirrhotic patients. Within 10 years after the diagnosis of compensated cirrhosis, about 50% of patients will have developed ascites. ­Approximately 15% of patients with cirrhosis die within 1 year of forming ascites, and 50% die within 5 years.

PATHOPHYSIOLOGY OF ASCITES

In patients with cirrhosis, progressive portal hypertension leads to the activation of endogenous vasoconstrictors and retention of sodium and water which then increases renal vasoconstriction. Cirrhotic vasodilation results in effective volume depletion. This is sensed as reduced pressure at the carotid and renal baroreceptors and activates sodium-retaining neurohumoral mechanisms. These mechanisms include the renin-angiotensin-aldosterone system, sympathetic nervous system and antidiuretic hormone. The net effect is avid sodium and water retention even though extracellular sodium stores and plasma volume are increased. Patients with ascites and urinary sodium below 10 mEq/d have a mean survival as low as 5 to 6 months in comparison to those who have a higher rate of sodium excretion. Similarly, the degree of water retention parallels the severity of cirrhosis. The degree of hyponatremia, as a result of increased antidiuretic hormone secretion, correlates with worsening survival. Over time, the expanded plasma volume and resulting lymph formation, in combination with decreased oncotic pressure from hypoalbuminemia, results in fluid retention within the peritoneal cavity.

EVALUATION AND DIAGNOSIS OF ASCITES

Physical exam

Patients with ascites typically report progressive abdominal distension that may be associated with abdominal discomfort, but may also be painless. The majority of patients with cirrhosis severe enough to cause ascites formation have stigmata of cirrhosis on physical examination, such as spider angioma, palmar erythema and abdominal wall collaterals.

Confirming ascites with physical examination can be challenging if the patient is obese or does not have a large amount of ascitic fluid. The most helpful physical sign is flank dullness, which is observed when there is at least 1500 mL of ascitic fluid in the abdomen. If no flank dullness is present, the patient has less than a 10% chance of having ascites. When flank dullness is observed, one should test for “shifting dullness,” which has 83% sensitivity and 56% specificity in detecting ascites. ­Examination of the jugular veins should be performed since an elevated jugular venous pressure suggests ascites from heart failure or constrictive pericarditis. An abdominal ultrasound confirms that ascitic fluid is present and can determine where a diagnostic paracentesis can be safely performed. Ultrasonography is also a safe and cost-effective modality to evaluate the liver parenchyma for evidence of cirrhosis or malignancy as well as to check the patency of the portal and hepatic veins.

Diagnostic paracentesis

A diagnostic paracentesis with appropriate ascitic fluid analysis is a quick and cost-effective method to determine the etiology of ascities and should be performed in the initial evaluation of all patients with new-onset or recurrent ascites or patients hospitalized with ascites. Infected ascitic fluid may cause or exacerbate complications of cirrhosis such as encephalopathy, fever, or renal insufficiency (Table 160-2). Abdominal paracentesis should not be delayed in the evaluation of patients with suspected spontaneous bacterial peritonitis (SBP) as it has been shown that mortality increases by 3.3% per hour of delay in performing paracentesis.

Complications of abdominal paracentesis are rare. The most common complication is leak of ascitic fluid, which may occur in up to 5% of patients. Leaks typically arise when a proper Z-track technique has not been used. More serious complications such as severe hemorrhage and infection have been found to occur less than 1% of the time. Death due to paracentesis is extremely rare.

The routine administration of blood products (fresh frozen plasma and/or platelets) in patients with cirrhosis is not recommended. Blood products, such as FFP and platelets, should be reserved for patients with clinically evident disseminated intravascular coagulation or clinically evident hyperfibrinolysis. Patients with hyperfibrinolysis may be treated with aminocaproic acid or intravenous tranexamic acid. In patients with disseminated intravascular coagulation, platelets, and in some cases, fresh frozen plasma may be given prior to paracentesis.

ASCITIC FLUID ANALYSIS

Initial tests that should be performed on the ascitic fluid include the gross appearance of the fluid, cell count and differential, total protein concentration, and albumin concentration. Serum albumin should be obtained so that a serum-to-ascites albumin gradient determination (SAAG) can be calculated. Culture of ascitic fluid should be sent in patients admitted to the hospital with ascites as well as patients with signs of decompensation such as fever, abdominal pain, azotemia, acidosis, or confusion. Many laboratories save an aliquot of fluid in case further testing is necessary.

The gross appearance of the ascitic fluid can range from clear fluid typically seen in the setting of cirrhosis, turbid or cloudy fluid in the setting of infection, milky fluid in the setting of chylous ascites, and bloody fluid in the setting of malignancy or a traumatic paracentesis.

The SAAG is the most helpful test to identify the presence of portal hypertension. The SAAG is calculated by subtracting the ascitic fluid albumin value from the serum albumin value, obtained on the same day. The presence of a gradient ≥1.1 g/dL indicates that the patient has portal hypertension as the cause of ascites with 97% accuracy. In patients with cirrhosis and ascites, the SAAG reflects the degree of sinusoidal hypertension as it is assumed that a SAAG greater than 1.1 is equal or exceeds a sinusoidal pressure of 12 mm Hg. The SAAG is not specific to ascites due to cirrhosis, but will be elevated with any disorder that leads to portal hypertension. Patients with portal hypertension plus a second cause of ascites will also have a SAAG ≥1.1 g/dL. Importantly, the SAAG will remain accurate despite fluid infusion and diuretic use.

The cell count with differential is the most helpful test performed on ascitic fluid to evaluate for infection. It should be ordered with every paracentesis, as ascitic fluid infection is a potentially reversible cause of death in patients with cirrhosis and ascites. The sample should be sent in a tube containing an anticoagulant to avoid clotting. The white blood cell and neutrophil count need to be corrected in patients with bloody samples. To obtain corrected white blood cell and neutrophil counts one white blood cell should be subtracted for every 750 red blood cells and one neutrophil should be subtracted for every 250 red blood cells. Antibiotic treatment should be considered in any patient with a corrected neutrophil count ≥250/mm³.

In cirrhosis total ascitic fluid protein is usually low (<1 g/dL) and high (>2.5 g/dL) in most other causes of ascites. For example, both ascites from cirrhosis and cardiac ascites have a SAAG ≥1.1 g/dL, but in cirrhosis the total protein is <2.5 g/dL, whereas in cardiac ascites it is ≥2.5 g/dL.

Milky ascitic fluid may suggest lymphatic obstruction and should be tested for elevated triglyceride levels. Cytology should be ordered when there is a high pretest probability of peritoneal carcinomatosis. Patients with peritoneal carcinomatosis usually have a history of breast, colon, gastric, or pancreatic carcinoma. The cancer antigen 125 (CA 125) is nonspecific in the differential diagnosis of ascites and is not recommended. In patients at high risk for tuberculous peritonitis, it is appropriate to check an adenosine deminase level and AFB culture (Table 160-3).

MANAGEMENT OF ASCITES

While there is no evidence that treatment of fluid overload in patients with cirrhosis improves survival, it has important benefits. Removing ascitic fluid volume provides important symptomatic relief to patients, including less abdominal discomfort and shortness of breath as well as improved mobility and appetite. Removal of fluid reduces the risk of cellulitis and abdominal wall hernia formation. It also concentrates ascitic fluid opsonins, which may reduce the risk of spontaneous bacterial peritonitis.

The treatment of ascites in patients with cirrhosis includes abstinence from alcohol, sodium restriction and diuretic therapy. In addition, patients with tense ascites should be treated with large-volume therapeutic paracenteses unless they have SBP, as it may precipitate renal injury. Removal of less than 5 L of fluid does not appear to have hemodynamic or hormonal consequences and is more rapid than diuretic therapy for ascitic fluid removal. For paracenteses over 5 L an albumin infusion of 6 to 8 g/L of fluid removed can be given and has been shown to have a survival advantage. The correct identification of the etiology of ascites is critical because patients with a low SAAG (<1.1 g/dL), such as ascites from peritoneal carcinomatosis, do not respond to sodium restriction and diuretic therapy.

The formation of ascitic fluid in a patient with portal hypertension is the result of avid renal retention of sodium and water. Therefore, a mainstay in the treatment of ascites is to restrict sodium intake. All patients with ascites should have sodium intake restricted to 88 mEq (2000 mg) per day. More stringent sodium restriction is not recommended because it is generally not tolerated, has poor compliance and may worsen malnutrition in patients with cirrhosis. ­Education about the importance of dietary sodium restriction is essential to the management of patients with portal hypertension related ascites as fluid loss and weight change are directly related to sodium balance.

Routine fluid restriction is not recommended in patients with cirrhosis and ascites. The chronic hyponatremia in patients with cirrhosis is rarely problematic and fluid restriction may worsen thirst and be uncomfortable for patients. Fluid restriction should be limited to patients with neurologic symptoms that may be due to severe hyponatremia (<120 mEq/L), which is an uncommon finding (1% in one series). The use of vaptans (vasopressin antagonists) in the treatment of severe hyponatremia in patients with cirrhosis is controversial and without clear evidence of a beneficial effect.

The majority of patients with cirrhosis and ascites will need diuretic therapy in addition to salt restriction to avoid fluid retention. The usual diuretic regimen consists of single morning doses of oral spironolactone and oral furosemide, which promote natruesis through different mechanisms. The starting doses are 100 mg of spironolactone and 40 mg of furosemide; this ratio generally maintains normokalemia. The doses of both oral diuretics can be increased every 3 to 5 days maintaining the 100 mg: 40 mg ratio to a maximum of 400 mg/d of spironolactone and 160 mg of furosemide. Diuretics should be initiated with a diuresis goal of 1 L/d. Spot urine sodium to potassium concentration greater than one indicates that a patient has effective aldosterone blockade and should respond to sodium restriction. All diuretics should be discontinued if there is severe hyponatremia (serum sodium concentration <120 mmol/L), progressive renal failure, worsening hepatic encephalopathy, or incapacitating muscle cramps. In addition, furosemide should be stopped if there is severe hypokalemia (<3 mmol/L) and aldosterone antagonists should be stopped if patients develop severe hyperkalemia (serum potassium >6 mmol/L). In hospitalized patients, diuretics should also be withheld in the setting active gastrointestinal bleeding, hepatic encephalopathy or renal dysfunction. Unlike patients with cardiac ascites, patients with cirrhotic ascites should not be treated with intravenous diuretics as this may lead to azotemia. The complete resolution of clinically apparent ascites is not a prerequisite for discharge.

Certain medications should be avoided or used with caution in patients with cirrhosis and ascites. In patients with cirrhosis arterial blood pressure independently predicts survival. Drugs that inhibit vasoconstrictors, such as angiotensin converting enzyme inhibitors and angiotensin receptor blockers should be avoided or used with caution because of the increased risk of renal impairment. Although propranolol has been shown to prevent variceal hemorrhage in patients with large varices, in patients with refractory ascites, β-blockers may adversely affect the patient. In patients with advanced cirrhosis, the risks versus benefits must be carefully weighed in each patient. Prostaglandin inhibitors, such as nonsteroidal anti-inflammatory drugs, should be avoided in patients since they can reduce urinary sodium excretion, lead to azotemia and may precipitate upper gastrointestinal bleeding in the setting of cirrhosis. In patients with cirrhosis and ascites, aminoglycosides alone or in combination with other antibiotics should be avoided in the treatment of bacterial infections because they are associated with a high incidence of nephrotoxicity. In addition, in hospitalized patients with cirrhosis and renal failure the use of contrast media should be used with caution.

REFRACTORY ASCITES

Approximately 10% of patients with ascites due to cirrhosis develop diuretic-resistant ascites, defined as refractory ascites. Refractory ascites in patients with cirrhosis is present when at least one of the following criteria is met: an inability to mobilize ascites despite confirmed adherence to the dietary sodium restriction and the administration of maximum tolerable doses of oral diuretics (400 mg/d of spironolactone and 160 mg/d of furosemide); rapid reaccumulation of fluid after therapeutic paracentesis; the development of diuretic-related complications, such as progressive azotemia, hepatic encephalopathy, or progressive electrolyte imbalances. In hospitalized patients spot urine sodium to potassium ratio is an efficient method to determine compliance and efficacy. A urine sodium greater than urine potassium with weight loss indicates the patient is diuretic responsive and adherent to the sodium restriction; a urine sodium less than urine potassium without weight loss is diuretic resistant at the current doses; a urine sodium greater than urine potassium without loss indicates that the patient is diuretic responsive, but not adherent to the sodium restriction. It is also critical to differentiate refractory ascites from causes of ascites such as malignant ascites or the Budd Chiari syndrome (hepatic vein thrombosis), which may be determined through ascitic fluid analysis and an abdominal ultrasound with Doppler.

The first step in the management of refractory ascites is to discontinue medications that decrease systemic blood pressure and lead to renal vasoconstriction, such as angiotensin converting enzyme inhibitors, angiotensin receptor II blockers, and nonsteroidal anti-inflammatory drugs. Although many patients with cirrhosis are on β-blockers, such as propranolol, to decrease the risk of variceal bleeding, the risks versus benefits of β-blockers must be weighed carefully in each patient. β-Blockers should be discontinued in patients with refractory ascites if the systemic blood pressure is less than 90 mm Hg, if hyponatremia (<130 mEq/L) is present, or if there is acute kidney injury. Oral midodrine, an oral vasopressor, which often increases blood pressure in advanced cirrhosis, may theoretically convert diuretic-resistant patients back to diuretic sensitive. Midodrine may be started at 5 mg orally three times daily with the dose adjusted by 2.5 mg every 24 hours up to a maximal dose of 17.5 mg to achieve a mean arterial pressure greater than 82 mm Hg.

Therapeutic options for patients who fail noninvasive treatments include serial therapeutic large-volume paracenteses (LVPs), transjugular intrahepatic portosystemic shunt, and liver transplantation. Liver transplantation is primarily based on the patient’s MELD (model for end stage liver disease) score but should not be delayed in patients with refractory ascites as 21% of patients with refractory ascites die within 6 months. Serial paracenteses of 4 to 6 L per session may be effective for symptomatic relief. The use of albumin replacement in LVP to prevent circulatory dysfunction after paracenteses is controversial. Based on the available data, for paracenteses larger than 5 L, albumin (6-8 g/L of fluid removed) may be administered either during or immediately after the procedure.

Transjugular intrahepatic portosystemic stent-shunt (TIPS) reduces portal hypertension and decreases the development of cirrhotic ascites. There is increasing evidence that TIPS is more effective than large-volume paracentesis in controlling ascites and may be associated with a survival advantage. As an invasive procedure, TIPS should only be considered in carefully selected patients with diuretic-resistant ascites intolerant of repeated large-volume paracentesis or patients requiring very frequent paracentesis (eg, weekly). Patients with refractory ascites who are not good candidates for TIPS include those with Child-Pugh class C cirrhosis, a high MELD score (>18), alcoholic hepatitis, congestive heart failure, a history of severe, spontaneous hepatic encephalopathy or the absence of a caregiver in the home. Echocardiograms are typically performed to screen for heart failure prior to TIPS placement. Patients with cirrhosis and ascites usually have ejection fractions of 70% to 75% due to hyperdynamic circulation. A baseline ejection fraction of 60% is a usual minimum cutoff prior to TIPS because central pressure usually increases after TIPS and cardiac function can deteriorate.

HEPATIC HYDROTHORAX

Hepatic hydrothorax is defined as the presence of a pleural effusion in a patient with cirrhosis and ascites in the absence of other causes. Hepatic hydrothorax occurs in approximately 5% to 10% of patients with cirrhosis and ascites. The development of hepatic hydrothorax results from the passage of ascites from the peritoneal cavity into the pleural cavity through small diaphragmatic defects during the negative intrathoracic pressure generated during inspiration. The right hemidiaphragm is more susceptible to diaphragmatic defects and hepatic hydrothorax develops on the right side in approximately 73% to 85% of patients. Hepatic hydrothorax usually presents with dyspnea, a nonproductive cough, pleuritic chest pain, and hypoxemia. Patients with suspected hepatic hydrothorax should undergo thoracentesis with testing of the pleural fluid as well as additional imaging including a chest computed tomographic scan and echocardiogram to confirm the diagnosis and exclude other causes of the effusion. Pleural effusions from hepatic hydrothorax are transudative in nature; however, the protein concentration of the pleural fluid is usually higher than that of ascitic fluid due to the differences in hydrostatic forces between the abdomen and chest. If the diagnosis is uncertain, an intraperitoneal injection of technetium-radiolabeled sulfur colloid into the abdomen can be performed shortly after therapeutic thoracentesis to detect rapid passage of isotope into the chest cavity during reaccumulation.

Patients who are severely symptomatic from a large effusion should undergo a therapeutic thoracentesis. Usually no more than 2 L of fluid are removed because of the risk of pulmonary edema and hypotension. After thoracentesis, patients should be on a sodium restricted diet (<88 mEq or 2000 mg daily) and diuretics to prevent reaccumulation of fluid. Chest tubes are contraindicated in the treatment of hepatic hydrothorax as they can result in massive protein and electrolyte depletion, infection, renal failure, and bleeding, and may result in rapid deterioration or death. Spontaneous bacterial empyema (SBEM), is defined as the lack of evidence of pneumonia on chest imaging study and a positive pleural fluid culture and a polymorphonuclear leukocyte count (PMN) cell count >250 cells/mm³ or negative pleural fluid culture and a PMN cell count >500 cells/mm³. Spontaneous bacterial empyema has been reported in 13% to 16% of patients with hepatic hydrothorax. Spontaneous bacterial empyema should be treated with intravenous antibiotics used for SBP (eg, ceftriaxone 2 g every 24 hours) and may be tailored based on culture results. A repeat thoracentesis may be performed to document a patient’s response to treatment.

Patients with refractory hydrothorax, defined as a persistent hydrothorax despite a sodium-restricted diet and diuretic therapy, may initially be managed with repeated thoracentesis every 2 to 3 weeks for symptom control. Transjugular intrahepatic portosystemic shunt is the second line treatment and may be recommended in patients with a Child-Pugh score <13 who are younger than 70 years old and do not have clinically significant hepatic encephalopathy. Pleurodesis or thorascopic repair of the diaphragmatic defects are technically difficult and have had variable results, but may be considered in patients with no other therapeutic options. Liver transplantation is the definitive treatment for appropriate candidates.

SPONTANEOUS BACTERIAL PERITONITIS (SBP)

Spontaneous bacterial peritonitis is defined by the presence of an elevated ascitic fluid absolute polymorphonuclear leukocyte count (PMN) count (>250 cells /mm³) or a positive culture without an evident intra-abdominal, surgically treatable source of infection. Spontaneous bacterial peritonitis often presents with fever, abdominal pain, or altered mental status, but patients can also be asymptomatic. A diagnostic paracentesis must be performed to diagnose or exclude SBP. Therefore, all patients hospitalized with ascites should undergo a diagnostic paracentesis. Patients should have a repeat paracentesis during their hospitalization if they develop clinical signs or symptoms suggesting infection. The paracentesis should be performed prior to the administration of any antibiotics. A single dose of a broad-spectrum antibiotic may inhibit culture growth in 86% of patients. Most cases of SBP have a single organism due to gut bacteria such as E. coli and Klebsiella, though streptococcal and staphylococcal infections may also occur. A positive ascitic fluid culture is not necessary for the diagnosis of SBP and culture negative, neutrophil positive ascites occurs commonly (15%-50%). Findings suggestive of a secondary peritonitis, such as perforated viscus, diverticulitis, or intra-abdominal abscess, include polymicrobial infection, anaerobes, high protein concentration (>1 g/dL), low glucose (<50 mg/dL) or high LDH (> serum upper limit of normal). Patients with secondary bacterial peritonitis should undergo abdominal computed tomography (CT) scanning and surgical evaluation as indicated.

Broad-spectrum therapy is warranted in patients with suspected SBP until the results of sensitivity testing has returned. The infection-related mortality from SBP in the absence of renal failure or shock is low, but up to 82% in patients with cirrhosis and SBP-associated septic shock. Antibiotics should be initiated immediately after diagnostic paracentesis, especially for patients who have developed septic shock. One series showed an adjusted odds ratio for mortality of 1.9 for every hour delay in administrating antimicrobial therapy. β-Blockers use among patients with SBP has been associated with increased risk of the hepatorenal syndrome and a probable increased mortality. Therefore, β-blockers should be discontinued in a hospitalized patient with SBP.

Uncomplicated SBP should be treated with a third-generation cephalosporin, such as cefotaxime. Intravenous cefotaxime 2 g every 8 hours produces excellent ascitic fluid penetration. Lower doses can be used in patients with impaired renal function. The main adverse drug reaction of cefotaxime is rash, which occurs in approximately 1% of patients. Levofloxacin may be used for patients with a penicillin allergy, although it has less penetration into ascitic fluid than cefotaxime. Oral ofloxacin (400 mg twice per day) may also be considered for SBP treatment in patients without shock, vomiting, or a serum creatinine greater than 3 mg/dL. Fluoroquinolones, however, should not be used in a patient who had been receiving a fluoroquinolone for SBP prophylaxis as resistance to the antibiotic may have developed. The choice of antibiotics for treating SBP should also take into account local resistance patterns and recent antibiotic use. Nephrotoxic antibiotics should be avoided because the underperfused kidneys in cirrhosis are very sensitive to injury.

In most patien.ts, including those who are bacteremic, 5 days of therapy is usually sufficient and antibiotics can be discontinued without a routine repeat paracentesis if the patient has clinically improved. However, if after 5 days of therapy, fever or pain persists, paracentesis is repeated, and the decision to continue or discontinue antibiotics is determined by the PMN response. If the PMN count is <250 cells/mm³, treatment is stopped; if the PMN count is elevated, but less than the pretreatment value, antibiotics are continued for another 48 hours, and paracentesis is repeated. If the PMN count is greater than the pretreatment value, a search for secondary bacterial peritonitis is initiated. Only patients who grow an unusual organism (eg, pseudomonas), an organism resistant to standard antibiotic therapy, or an organism routinely associated with endocarditis (eg, Staphylococcus aureus or viridans group streptococci) are initially considered for longer treatment.

Renal failure develops in 30% to 40% of patients with SBP and is a major cause of death. This risk may be reduced with an infusion of intravenous albumin (1.5 g/kg body weight within 6 hours of diagnosis and 1.0 g/kg on day 3). Albumin infusion should be given if the creatinine is >1 mg/dL, the blood urea nitrogen is >30 mg/dL, or if the total bilirubin is >4 mg/dL.

It is reasonable to give empiric therapy to patients with alcoholic hepatitis who present with fever and/or leukocytosis, but have a PMN count <250 cells/mm³. Empiric therapy can be discontinued after 48 hours if ascitic fluid, blood, and urine cultures remain negative.

Recurrence of SBP has been reported to be 69% in 1 year. Risk factors for the development of SBP include ascitic fluid protein concentration <1.0, variceal hemorrhage, or a prior episode of SBP. The use of antibiotic prophylaxis for patients at high risk for SBP decreases the risk of bacterial infection and mortality. However, the use of antibiotic prophylaxis does select for resistant flora, which can result in recurrent infection. Therefore, antibiotic prophylaxis should be limited to patients with strict indications. These include an ascitic fluid total protein less than 1.5 g/dL with impaired renal function (creatinine ≥1.2, BUN ≥25 or serum Na ≤130) or liver failure (Child score ≥9 and bilirubin ≥3), patients with a previous episode of SBP, or in patients with cirrhosis and gastrointestinal bleeding. The antibiotic regimen used for SBP prophylaxis depends on the indication. For patients with an ascitic fluid total protein less than 1.5 g/dL and with impaired renal function (creatinine ≥1.2, BUN ≥25 or serum Na ≤130) or liver failure (Child score ≥9 and bilirubin ≥3) or patients with a previous episode of SBP, prolonged therapy with norfloxacin 400 mg/d, orally, is preferred. Ciprofloxacin 750 mg/wk or trimethoprim-sulfamethoxazole (800 mg sulfamethoxazole and 160 mg trimethoprim, daily, orally) are acceptable alternatives if norfloxacin is not tolerated. In patients with cirrhosis and gastrointestinal bleeding, intravenous ceftriaxone 1 g daily is initiated and can be switched to oral norfloxacin (400 mg orally twice daily) for a total course of 7 days once bleeding has been controlled and the patient is stable.

In addition to antibiotic prophylaxis, general measures may be taken to prevent SBP in hospitalized patients with cirrhosis and ascites. The early recognition and aggressive treatment of localized infections (eg, cystitis and cellulitis) as well as the judicious use of urinary catheters can prevent bacteremia and SBP. Diuretic therapy, if not contraindicated, concentrates ascitic fluid and raises opsonic activity, which may help prevent SBP. Protein pump inhibitors have been associated with an increased risk of SBP and should be restricted to patients with a clear indication for their use (­Figure 160-3).

HEPATORENAL SYNDROME

The hepatorenal syndrome is defined as the occurrence of renal failure in a patient with cirrhosis and ascites in the absence of other causes of renal failure, shock or hypovolemia. Hypovolemia is excluded by the absence of response defined as no sustained improvement of renal function (creatinine decreasing to <133 μmol/L) following at least 2 days of volume expansion with intravenous albumin (1 g/kg of body weight per day up to 100 g/d) and/or the discontinuation of diuretics. Other apparent causes for acute kidney injury should be excluded including current or recent treatment with nephrotoxic drugs, and the ultrasonographic evidence of obstruction or parenchymal disease. A kidney biopsy is often helpful in excluding pre-existing renal disease, especially if liver transplantation is being considered. The diagnosis is based on an increase in serum creatinine of 0.3 mg/dL or more within 48 hours, or an increase from baseline of 50% or more within 7 days. In hospitalized patients, repeat measurements of creatinine are helpful in the early diagnosis of hepatic renal syndrome (HRS). ­Additional clinical criteria include urine red cell excretion of less than 50 cells per high power field (in the absence of a urinary catheter) and protein excretion less than 500 mg/d The hepatorenal syndrome occurs in approximately 20% of patients with cirrhosis and ascites at 1 year and up to 40% at 5 years and is associated with a high mortality rate. There are two forms of hepatorenal syndrome based on the rate of decline in renal function. Type 1 hepatorenal syndrome is the more serious form and is defined as at least a twofold increase in serum creatinine to a level greater than 2.5 mg/dL during a period of less than 2 weeks. Precipitating conditions include infection, especially SBP, and severe alcoholic hepatitis. The median survival for untreated type 1 HRS is 1 month. The early diagnosis and treatment of infection can reduce the risk of HRS and is associated with an improved survival. Type 2 hepatorenal syndrome is defined as renal impairment that is less severe than in patients with the type 1 form. It is important to note that diuretics do not cause hepatorenal syndrome but can lead to diuretic-induced azotemia, which improves with the cessation of therapy and fluid repletion. Patients with relative adrenal insufficiency have greater impairment of circulatory and renal function and a higher probability of type 1 HRS and severe sepsis when compared to patients with decompensated cirrhosis but normal adrenal function.

TREATMENT OF HRS

The goal of medical therapy is to improve renal function by reversing portal hypertension induced arterial vasodilation in the splanchnic circulation. In patients with HRS who are admitted to the intensive care unit, careful monitoring of urine output, fluid balance, arterial pressure and vital signs is imperative. Initially, all diuretics should be discontinued. In patients with tense ascites, paracentesis may be helpful in alleviating abdominal discomfort.

Medical therapy is aimed at improving the severely impaired circulatory function by inducing vasoconstriction of the dilated vascular bed and increasing arterial pressure. Based on results of many randomized controlled trials, the use of vasoconstrictive drugs, especially terlipressin, has been associated with improvement in up to 50% of patients. Terlipressin in combination with albumin is the treatment of choice. Terlipressin is given as an intravenous bolus of 1 to 2 mg every 4 to 6 hours, and albumin is given for 2 days as an intravenous bolus of 1 g/kg/d (100 g maximum), followed by 25 to 50 g/d until terlipressin therapy is discontinued. ­Contraindications to terlipressin therapy include ischemic cardiovascular diseases. Patients treated with terlipressin should be carefully monitored for development of cardiac arrhythmias or signs of splanchnic or digital ischemia. Terlipressin is currently unavailable in the United States. Alternative treatments include noradrenaline and midodrine, a selective alpha-1 adrenergic agonist, plus octreotide, a somatostatin analog, in addition to albumin infusion. Midodrine is given orally at a starting dose 7.5 mg three times daily, uptitrated to 15 mg three times per day. The midodrine dose should be uptitrated with each consecutive dose to raise the mean arterial pressure by approximately 10 to 15 mm Hg. Octreotide is either given as a continuous intravenous infusion of 50 μg/h or subcutaneously 100 to 200 mg three times daily. Albumin is given for 2 days as an intravenous bolus of 1 g/kg (100 g maximum), followed by 25 to 50 g/d until midodrine and octreotide therapy is discontinued. Norepinephrine is administered as a continuous infusion (0.5-3 mg/h) with the goal of raising the mean arterial pressure by 10 to 15 mm Hg.

Patients treated with norepinephrine, terlipressin, or octreotide are typically treated for 2 weeks. The duration may be extended, however, if after 2 weeks there is some, but not complete improvement in renal function. In patients who respond to therapy, treatment with oral midodrine to maintain higher mean arterial pressure can be continued at discharge until resolution of liver injury or liver transplantation in appropriate candidates. If a patient has no improvement in renal function after 2 weeks then medical therapy is discontinued as the outcome is usually futile. In a recent study, patients with type 1 HRS associated with infection, HRS was not reversible despite treatment with vasoconstrictive drugs in two thirds of patients. Therefore, these patients must be treated immediately with antibiotics if there is suspicion of infection.

Options for patients who fail to respond to medical therapy include TIPS and hemodialysis. The use of TIPS in patients with hepatorenal syndrome may provide short-term benefit in selected patients but data to its applicability is very limited and many of these patients have contraindications to TIPS placement. Given the risks associated with the procedure, TIPS should be limited to patients who are well enough to tolerate the procedure and are awaiting liver transplantation. In patients who fail medical therapy and are not considered candidates for TIPS, hemodialysis or continuous venous hemofiltration can be considered as treatment for severe metabolic abnormalities such as metabolic acidosis and hyperkalemia but there is very limited data as to their efficacy. Studies are currently underway investigating artificial liver support systems in the management of HRS. In appropriate candidates, liver transplantation is the treatment of choice for both type 1 and type 2 HRS, with survival rates of approximately 65% in type 1 HRS.

HEPATIC ENCEPHALOPATHY

Hepatic encephalopathy represents a spectrum of neuropsychiatric disturbances in patients with impaired hepatic function. In patients with cirrhosis, overt hepatic encephalopathy develops in 30% to 45% of patients. Neuropsychiatric findings range from subtle abnormalities that are not apparent without specialized testing to overt coma. Important factors that determine the severity of hepatic encephalopathy include level of consciousness, cognitive function, behavioral disturbances, and neuromuscular features.

Although the exact pathogenesis of hepatic encephalopathy remains unclear, most theories suggest that the brain is exposed to toxic substances that are produced by bacterial flora, which are incompletely cleared by the compromised liver. The most well described neurotoxic substance is ammonia. The majority of patients with overt hepatic encephalopathy will have elevated arterial blood levels of ammonia, though the degree of ammonia elevation does not necessarily correlate with degree of hepatic encephalopathy. Gamma-aminobutyric acid (GABA), the primary inhibitory neurotransmitter in the central nervous system, may also play an important role in the development of hepatic encephalopathy. It has been theorized that gut bacteria produce increased endogenous benzodiazepine ligands that bind the benzodiazepine site of GABA receptors.

The generally accepted classification for hepatic encephalopathy takes into account the type of hepatic abnormality and the duration and characteristics of the neuropsychiatric manifestations. Hepatic encephalopathy may be associated with acute liver failure (type A), portosystemic shunting without intrinsic liver disease (type B), and cirrhosis (type C). Hepatic encephalopathy associated with cirrhosis is characterized as episodic, recurrent, persistent, or minimal. Episodic hepatic encephalopathy develops over a short period of time and is further classified based on whether it was precipitated or has developed spontaneously. Recurrent encephalopathy is defined as bouts of hepatic encephalopathy that occur within a time interval of 6 months or less. Patients with persistent encephalopathy exhibit continuous overt neurologic or behavioral abnormalities, which are interspersed with episodes of overt hepatic encephalopathy. Patients with minimal hepatic encephalopathy often appear asymptomatic and have subtle neuropsychiatric abnormalities that may only be detected with psychomotor or electrophysiologic testing. Patients with overt hepatic encephalopathy have signs and symptoms that can be detected clinically. The severity of overt hepatic encephalopathy is graded from I to IV. Grade I: patients have changes in behavior, mild confusion, slurred speech, and disordered sleep (both hypersomnia and insomnia); Grade II: patients have increasing lethargy and moderate confusion; Grade III: patients have marked confusion (stupor), incoherent speech, and often sleeping yet arousable; Grade IV: patients are in a coma and unresponsive to pain. In reality, it is difficult to differentiate between Grade II and Grade III hepatic encephalopathy (Table 160-4).