Introduction

Upper gastrointestinal (GI) bleeding is a common GI emergency. Frequency of occurrence is 40 to 150 cases per 100,000 in the Western population, and it accounts for a total expenditure of $2.5 billion annually in the United States.^1,2 Significant advances have been made not only in decreasing the overall incidence, but also in treatment (medical treatment and endoscopic hemostasis), prevention of complications, and recurrence of bleeding.³ These advances have resulted in an overall decrease in the rate of hospitalizations and mortality associated with upper GI hemorrhage in the Western world.⁴ However, more recent data suggest that although hospital admission rates may have decreased, patients being admitted are older and have more comorbid illnesses. These patients are known to carry a worse prognosis.³ Acute nonvariceal upper GI bleeding is still associated with significant morbidity and mortality despite significant development in the understanding of the pathophysiology of the disease and the armamentarium available to the endoscopist for endoscopic therapy to treat these lesions. This chapter discusses the approach, assessment, and management strategies in patients presenting with nonvariceal upper GI bleeding.

Clinical Presentation

The presentation of GI bleeding depends on the volume and site of bleeding. Hematemesis is vomiting of blood and is the most common presentation of an upper GI bleed. The source is almost always proximal to the ligament of Treitz. Blood may be bright red, indicating a recent bleed, or resemble “coffee grounds,” representing older blood reduced by acid in the stomach. Melena is black, tarry, and sticky stool with a specific foul odor caused by degradation of blood in the intestines and colon. This presentation most commonly indicates an upper GI source; however, blood from the right colon may also manifest as melena. A massive upper GI bleed can manifest as hematochezia (bright red blood per rectum) in 15% of cases and carries a worse prognosis.⁵

Initial Evaluation

The initial assessment of a patient suspected to have an acute GI bleed must focus on hemodynamic stability so that resuscitative measures can be started without delay. Good clinical assessment determines the fluid requirement for hemodynamic stabilization.⁶ A focused history and physical examination provides vital information on the severity of bleeding and other confounding medical problems (e.g., coronary disease, chronic obstructive pulmonary disease, malignancy) that may affect medical management and therapeutic intervention. The initial evaluation should focus on vital signs and orthostatic changes because postural hypotension represents a significant volume loss (>15%) and is predictive of a poor outcome (Table 12.1).^7,8

Peripheral intravenous (IV) access with two large-bore (at least 18-gauge) catheters or central venous access must be achieved in patients with an acute bleed, especially if the patient is hemodynamically unstable. Crystalloids (normal saline or lactated Ringer’s solution) are the initial fluids of choice and are administered rapidly to restore intravascular volume to ensure adequate tissue perfusion and oxygen delivery. Continuous monitoring is required, especially for fragile patients with comorbid cardiopulmonary disease.

Nasogastric Aspirate

Nasogastric tubes can be helpful in the localization of bleeding because a positive nasogastric aspirate (coffee ground or bright red blood) confirms the source to be from the upper GI tract. However, the localization of bleeding can be determined by a careful history and physical examination. A reliable history of hematemesis confirms the source to be proximal to the ligament of Treitz. Use of a nasogastric tube to stratify patients with high-risk lesions is controversial.⁹ In the national survey of the American Society of Gastrointestinal Endoscopy (ASGE), about 16% of patients with clear nasogastric aspirates were found to have active bleeding at the time of endoscopy.¹⁰ The bleeding sources in these patients also included esophagitis (10.7%) and varices (5.1%).

Aljebreen and colleagues¹¹ performed a retrospective analysis on 1869 patients with upper GI bleeding. A bloody nasogastric aspirate was significantly associated with high-risk lesions (active bleeding, nonbleeding visible vessel, adherent clot). A bloody aspirate had a specificity of 75.8% and negative predictive value of 77.9% for high-risk lesions. Although nasogastric lavage provides no information about the etiology of bleeding, it can be valuable in localizing the source in a hemodynamically unstable patient because it is a quick and easy test that is performed at bedside. A negative or bilious aspirate does not rule out upper GI bleeding. Nasogastric suction often is not needed with the availability of large-channel therapeutic endoscopes, with forward water jets built into the endoscope and the ability to suction directly from the endoscopic channel port, bypassing the internal suction channeling that passes through the handle of the endoscope and through the umbilicord, which can contain a stepdown in channel size.

Laboratory Data

Laboratory tests appropriate at initial presentation include hemoglobin level, hematocrit, platelet count, prothrombin time, and partial thromboplastin time. Initial hemoglobin level may not depict the degree of bleeding, and initial decision making must be based on clinical grounds. An increased blood urea nitrogen (BUN)-to-creatinine ratio (>36) has been suggestive of an upper GI source of bleeding with a sensitivity of 90% and specificity of 27%.¹² BUN levels can weakly predict severity of bleeding but are not helpful in predicting high-risk lesions.¹³ BUN values can be helpful in the diagnosis; however, the clinical picture is often complicated by other medical illnesses (e.g., renal insufficiency, congestive heart failure) and polypharmacy. Other important laboratory data include liver function tests and cardiac enzyme analysis.

Risk Stratification

Most GI bleeds stop spontaneously without any recurrence. Approximately 20% of bleeds can continue or recur leading to patient morbidity and mortality.¹⁴ Numerous clinical factors have been identified that predict a high rate of recurrent bleeding or poor outcome. Old age (>65 years), shock, comorbid illnesses, low hemoglobin on evaluation, melena, multiple transfusions (more than five), hematochezia, fresh blood emesis or nasogastric aspirate, and need for emergency surgery are clinical predictors of increased risk of rebleeding.^15,16 A history of alcoholism, a history of cancer, and an Acute Physiology, Age, and Chronic Health Evaluation (APACHE) II score of 11 or greater have also been associated with poor outcomes.^17,18

Chiu and colleagues¹⁹ prospectively studied 3220 patients with bleeding peptic ulcers from 1993 to 2003 and identified risk factors for mortality. Among patients with bleeding peptic ulcers after endoscopic hemostasis, advanced age, multiple comorbidities, hypovolemic shock, in-hospital bleeding, rebleeding, and need for surgery were significant factors predicting in-hospital mortality (Table 12.2). Based on these individual clinical criteria and endoscopic findings, numerous scoring systems have been formulated to try to stratify high-risk patients for appropriate intervention.

Table 12.2 Statistically Significant Predictors of Persistent or Recurrent Bleeding

NA, Not available.

Modified from Barkun A, Bardou M, Marshall JK; Nonvariceal Upper GI Bleeding Consensus Conference Group: Consensus recommendations for managing patients with non-variceal upper gastrointestinal bleeding. Ann Intern Med 139:843–859, 2003.

Blatchford and associates²⁰ devised a scoring system based on admission hemoglobin, BUN, pulse, systolic blood pressure, presentation with syncope and melena, and evidence of hepatic disease and cardiac failure. Their criteria could identify 20% of patients with very low risk of needing treatment to control bleeding, who could be offered outpatient treatment. The Blatchford scoring system has been validated by other investigators.²¹ In 1974, Forrest and colleagues²³ first classified stigmata of active or recent bleeding based on visualization at the time of endoscopy (Table 12.3). Visible vessel is the term used for an elevated area within the ulcer base. It is thought to be a coagulum over a small-caliber arterial vessel in the ulcer base.²⁴ The color of the lesion can also be predictive of rebleeding: Nonpigmented lesions (white-pale) have a higher risk (71%) than pigmented lesions (38%).²⁵ The frequency with which a visible vessel can be found also depends on the timing of endoscopy and how aggressively the clot is washed to expose the ulcer base. Chung and coworkers²⁶ reported disappearance of visible vessels in 62 patients who underwent endoscopy for 3 consecutive days.

Table 12.3 Modified Forrest Criteria

Rockall and colleagues²⁷ developed a scoring system involving clinical and endoscopic criteria. The main aim is to predict risk of rebleeding and mortality. An initial calculation is based on clinical parameters, and further categorization of the bleeding lesion and the stigmata of hemorrhage at the time of endoscopy enables the complete score to be calculated. A total score of less than 3 was associated with excellent prognosis, whereas a score greater than 8 carried a high mortality. For cases with a score less than 3, rebleeding occurred in less than 5%, and mortality was 0%. Multiple studies have been performed to validate the Rockall scoring system.^28,29 It has been validated in other studies to be predictive and cost-effective (Table 12.4).³⁰

Hay and coworkers³¹ derived a scoring system by using hemodynamics, time from bleeding, comorbidity, and endoscopy findings. Use of the practice guidelines reduced hospital stay in low-risk patients from 4.6 to 2.9 days (P < .001). Saeed and associates³² proposed another scoring system based on endoscopy findings. In a retrospective analysis, Chuan and colleagues compared the Blatchford score with the Rockall score in 354 patients. The Blatchford score identified 92.1% of patients, whereas the Rockall score identified 70.1%. Blatchford scoring based on clinical and laboratory parameters was concluded to have a higher sensitivity to identify high-risk patients (P < .0001).^33,34

Kim and associates³⁵ conducted a prospective study to compare the clinical utility of five scoring systems for the prediction of rebleeding and death in patients with nonvariceal upper GI bleeding. Using five scoring systems (Forrest classification, Rockall scoring system, Cedars-Sinai Medical Center Predict Index, Blatchford scoring system, and Baylor College scoring system), 239 consecutive patients were investigated. All patients underwent endoscopy for nonvariceal bleed. There were 35 patients (14.6%) who experienced rebleeding and 20 patients (8.4%) who died. Forrest classification was superior to the others in predicting rebleeding and death. The Cedars-Sinai Medical Center Predict Index and the Rockall scoring system showed high positive predictive values for predicting rebleeding (Cedars-Sinai Medical Center Predict Index) and death (Rockall scoring system). Forrest classification was thought to be the most useful scoring system for the prediction of rebleeding and death in patients with nonvariceal upper GI bleeding. The development of clinical scoring systems is very encouraging because these can be used by emergency department physicians, general practitioners, and junior house staff to triage patients effectively. These scoring systems have been validated in large cohorts in diverse populations and have been shown to be effective in risk stratification and cost-effectiveness.

Timing of Endoscopy

Endoscopy primarily is used to determine the cause of bleeding, prognosticate, and administer appropriate endoscopic therapy to control bleeding and prevent rebleeding.^36,37 The timing of endoscopy has been a subject of debate, especially in patients who are clinically stable with no evidence of further bleeding. It is now evident that early endoscopy provides essential data affecting patient triage.

Lee and associates³⁸ showed that early “endoscopy triage” leads to early discharge of patients with low risk of rebleeding, without increasing morbidity and mortality. Median cost savings were $2068. Cipolletta and coworkers³⁹ used endoscopic and clinical criteria to identify patients at low risk for recurrent bleeding. Patients were randomly assigned to outpatient care versus hospital admission. No patients underwent surgery or died. Rates of recurrent bleeding were comparable (2.1% and 2.2%) in both groups. Median costs were $340 for outpatients and $3940 for inpatients. Based on multiple studies confirming the beneficial effects of endoscopic therapy, urgent endoscopy has been recommended by the National Institutes of Health⁴⁰ and ASGE⁴¹ for patients who have active bleeding or who are at high risk for rebleeding. The definition of urgent endoscopy ranges in various studies from 2 to 24 hours after presentation to the hospital. It has been determined that 76% to 78% of patients with acute GI bleeding undergo endoscopy within the first 24 hours.^42,43 Even if discharge is not always the best option, triage results might help choose the correct level of care within the hospital.

Preparation and Place for Endoscopy

Endoscopy is best performed in a fully equipped endoscopy suite where staff members are trained to take care of the patient and in the use and maintenance of endoscopes and their accessories (Fig. 12.1). A general recommendation is that patients with mild to moderate bleeding can have endoscopy the next day in the endoscopy unit. Patients whose presentation suggests a major or severe bleeding event (e.g., hemodynamic compromise) can undergo endoscopy at the bedside in the intensive care unit after initial resuscitation. Endotracheal intubation to protect the airway and prevent aspiration should be considered in high-risk patients who are experiencing active hematemesis, marked agitation, lethargy, and persistent shock. In the United States, there is increasing use of propofol for routine and even emergency endoscopy; use of propofol not administered by an anesthetist has also been studied but is not widely employed in current practice.⁴⁴

Fig. 12.1 A, Comparison of the tips of endoscopes: diagnostic (tip diameter approximately 10 mm, channel size 2.8 mm); dual-channel therapeutic (tip diameter approximately 12 mm, channel sizes 3.4 mm and 2.8 mm); jumbo-channel therapeutic (tip diameter approximately 13 mm, channel size 6.0 mm). B, Jumbo-channel endoscope uses the biopsy channel port as the site for a separate suction line. A specialized toggle-type switch allows the suction to be controlled as desired and maintain use of the channel for devices. Use of the biopsy port with a second suction line is a practical method to deal with quick evacuation of blood and clot with any type of endoscope.

In patients with active or recent hemorrhage, residual blood in the stomach, especially in the fundus, can limit the quality of the examination, and residual clots must be removed or flushed away. Poor visualization at the time of endoscopy has been associated with worse outcomes.⁴⁵ Multiple methods have been proposed to overcome this problem. Vigorous gastric lavage using a large-bore nasogastric tube at the time of endoscopy and instillation of 3% hydrogen peroxide⁴⁶ to dissolve small clots have been proposed. Use of an endoscope with a large device channel may offer a convenient and effective method to remove clots and blood from the stomach.^47,48 Connecting a suction line directly to the endoscope device channel, bypassing the internal system, provides rapid and impressive clearing of blood and clot. Randomized controlled trials have looked into using erythromycin as a promotility agent to improve quality and yield of endoscopy.^49–52 Patients with acute GI bleeding were randomly assigned to receive erythromycin (3 mg/kg intravenously) and were compared with a no treatment arm. The treatment group (erythromycin 3 mg/kg intravenously 20 to 60 minutes or 120 minutes before endoscopy) had an improved quality of endoscopic examination and reduced need for second-look endoscopy, and erythromycin was deemed a cost-effective strategy.

Etiology of Nonvariceal Upper Gastrointestinal Bleeding

Causes of nonvariceal upper GI bleeding are presented in Table 12.5.

Table 12.5 American Society of Gastrointestinal Endoscopy Bleeding Survey: Endoscopic Diagnosis for Upper Gastrointestinal Bleeding in 2225 Patients

Available Modalities for Endoscopic Therapy

Endoscopic therapy is now an accepted community standard of care in patients with acute GI bleeding, with earlier defined risk factors and stigmata for rebleeding. Endoscopic therapy can be broadly divided into four categories: injection therapy, thermal coaptive therapy, mechanical devices, and combination methods. Combination therapies are more popular with data supporting this method of therapy in specific situations. Data are also supportive of thermal or mechanical therapy over injection therapy alone. For this reason, monotherapy with injection cannot be supported.

Thermal Coaptive Therapy

Thermal methods formerly included the neodymium:yttrium-aluminum-garnet (Nd:YAG) laser and presently include heater probe and electrocoagulation (Fig. 12.2). Laser therapy and monopolar coagulation are no longer popular because of the lack of portability to bedside, high cost, and perforation risks. Multipolar electrocoagulation and heater probe are the most widely used thermal modalities. The advantages of these devices are their excellent efficacy; safety; portability; and ability to combine irrigation, tamponade, and coagulation.

Fig. 12.2 Coagulation probes (from top down): argon plasma coagulation end-firing probe, Gold Probe (Boston Scientific, Microvasive Endoscopy, Natick, MA), and heater probe (Olympus America, Center Valley, PA).

In coaptive coagulation, the probe is used to compress physically and tamponade the bleeding vessel, followed by thermal energy sealing the walls of the vessel. In an animal model, arteries with a diameter of 2.5 mm can be coagulated with a heater probe using this technique.⁹³

Technique

Bipolar Electrocautery (Multipolar Probe)

The technique for applying bipolar electrocoagulating energy has not been standardized and varies among reported clinical trials. In canine models, a large (3.2-mm) multipolar probe produced better hemostasis than a smaller (2.3-mm) probe.^94,95 Laine⁹⁶ recommends forcefully applying a large (3.2-mm) bipolar circumactive probe (BICAP) on a power setting of 3 to 5 for a prolonged duration, such as 14 seconds, or seven pulses of 2 seconds each. Jensen and Hirabayashi⁹⁴ used a setting of 3 to 4 with 10-second pulses on a BICAP II generator. The Gold Probe (Boston Scientific, Microvasive Endoscopy, Natick, MA), which contains a gold foil–wrapped tip, has been shown in clinical trials to be effective at low power settings with longer pulse duration.⁹⁷

Heater Probe

The present technique for bleeding peptic ulcers involves using the larger heater probe, firm tamponade directly on the bleeding point or visible vessel, and coagulation with at least 120 J (four pulses of 30 J each) before moving the probe.⁹⁸ An additional application of energy when using the heater probe or multipolar (bipolar) probes as the probe is removed from the treatment site can reduce unwanted tissue adhesion and the inadvertent tearing of the coagulated sealant with resultant rebleeding. In a prospective study by Bianco and colleagues,⁹⁹ combined therapy with epinephrine injection with bipolar probe showed better results than epinephrine alone.

Mechanical Devices

Numerous mechanical devices, including hemoclip, endoloop, and rubber band, are now available for endoscopic treatment of GI bleeding. Although endoloop and rubber band ligation have little role in treating peptic ulcer bleeding, use of hemoclips has become very popular for emergency nonvariceal hemostasis. The underlying mechanism of action is mechanical clamping of the bleeding vessel. Various clips are available (Table 12.6). Multiple randomized trials have compared use of clips alone or in combination with other therapies. Most studies have been performed on first-generation clips. Jensen and associates¹⁰⁰ studied hemostasis capability of three presently available clips in a canine model; initial hemostasis was 100% for all types with longer retention times reported for the largest of the available clips, which can also be opened and closed on demand before deployment (see Table 12.6).

Table 12.6 Single-Use, Disposable Clipping Devices

Clips are easier to apply with the scope in straight position. Tangential access makes application a challenge because of the simple mechanics allowing the clip to be extruded from the catheter, opened, and finally closed. Similarly, it is difficult to place a clip in the fundic region with the scope retroflexed. All currently available clips are not designed for use through a side-viewing duodenoscope. The angle of entry of the clipping device and the elevator function challenge current clip designs. Various maneuvers and modifications of the actual device have been identified to allow use of the device in conjunction with a duodenoscope.

Among trials comparing hemoclip with other endoscopic therapy, Cipolletta and colleagues¹⁰¹ randomly assigned 112 patients with major stigmata to heater probe or hemoclip therapy (Fig. 12-3). Rates of recurrent bleeding (21% vs. 1.8%) and need for surgery (7% vs. 3.6%) were significantly lower in the hemoclip group. The investigators concluded that hemoclip therapy was safe and superior to heater probe therapy. In a randomized study by Chou and coworkers,¹⁰² hemoclip therapy was found to be superior to distilled water injection therapy in patients with major stigmata (active bleeding vessel, nonbleeding visible vessel). In their study comparing hemoclip versus heater probe, Lin and associates¹⁰³ reported 85% initial hemostasis in the hemoclip group compared with 100% in the heater probe group. Rebleeding rate was 8.8% in the hemoclip group and 5% in the heater probe group. In trials comparing combination therapy, Chung and colleagues¹⁰⁴ prospectively assigned 124 patients to hemoclip, hypertonic saline–epinephrine (HSE), and combined treatment groups. Initial hemostasis was comparable in all groups. Rates of recurrent bleed in hemoclip, HSE, and combination therapy groups were 2.4%, 14.6%, and 9.5%. Buffoli and colleagues¹⁰⁵ did not find an additional advantage of hemoclip therapy when used in combination with epinephrine injection therapy. A favored trend toward reducing surgery was seen in the combination therapy group (0% vs. 7.4%).

Fig. 12.3 Original manually loaded hemoclip (Olympus America, Center Valley, PA).

Several randomized controlled trials investigating the use of endoscopic clips alone or in combination with other endoscopic modalities have reported variable success (Table 12.7).¹⁰⁶ The most important factor seems to be difficulty in appropriate placement of the clip for effective hemostasis, especially for ulcers in difficult-to-approach sites such as high on the lesser curvature and posterior duodenal wall.¹⁰³ Improvements in endoscopic clip devices have included single-use, rotatable, and reopening features. Present data suggest clips to be as effective as thermal therapy. The choice between the two modalities is at the discretion of the endoscopist depending on his or her comfort level with the device and the location of the lesion (see Table 12.7).¹⁰⁶

Table 12.7 Randomized Controlled Trials of Clipping for Bleeding Peptic Ulcers

Combination Therapy

There is a theoretical advantage of combining therapies because hemostasis is achieved by different mechanisms. The most popular combination therapy used is injection therapy and thermal coagulation (Fig. 12.4); epinephrine causes vasoconstriction, reduced blood flow, and platelet activation, which potentiates the coaptive coagulation by thermal energy. Another advantage may be improved visualization of the target area for thermal coagulation after initial injection therapy in an actively bleeding ulcer. Combination therapy with injection and coaptive therapy has consistently shown superiority over medical therapy.¹¹⁵

Fig. 12.4 Large-sized heater probe and large-sized visible vessel protruding from an antral ulcer (not seen). Note the surrounding mucosal pallor from prior injection with 1 : 10,000 epinephrine.

In a randomized controlled trial, Chung and colleagues¹¹⁶ compared epinephrine injection with epinephrine injection plus heater probe in actively bleeding ulcers. Patients with major, active arterial bleeding had a better outcome regarding need for surgery (29.6% vs. 6.5%) and length of hospital stay (6 days vs. 4 days) after combination therapy. Combination therapy did not improve outcomes of patients with lesions with oozing only. Lin and coworkers¹¹⁷ compared epinephrine injection alone, bipolar electrocoagulation alone, and combined treatment. Combination therapy was associated with reduced rate of rebleeding, reduced treatment failure, and reduced transfusion requirements. In contrast, Marmo and associates¹¹⁸ showed single endoscopic therapy with thermal probes or clips is as effective as combination therapy and may be safer than combination therapy including injection therapy in high-risk ulcers.

A meta-analysis performed by Calvet and colleagues¹¹⁹ looked at trials comparing epinephrine injection alone versus epinephrine plus a second method (thermal, mechanical, sclerosant) for patients with upper GI hemorrhage from peptic ulcer disease. The analysis included 16 studies with 1673 patients; only studies with major stigmata of bleeding defined by Forrest criteria IA, IIA, and IIB were included. Addition of a second endoscopic method reduced rate of rebleeding (18.4% vs. 10.6%), need for surgery (11.3% vs. 7.6%), and mortality (5.1% vs. 2.6%) compared with epinephrine injection alone. Risk of further bleeding decreased regardless of the type of second method used. Risk of significant complications, including massive bleeding, gastric wall necrosis, and perforation, was the same in both groups (1.1%).

A meta-analysis by Yuan and associates¹²⁰ comparing clips with other modalities showed that initial hemostasis was 92% for clips versus 96% for other modalities (CI 0.19 to 1.75); however, rate of rebleeding was 8.5% for clips versus 15.5% for other modalities (CI 0.30 to 1.05). Both values did not reach statistical significance. Hemoclips are equivalent to other endoscopic modalities in terms of initial hemostasis, rebleeding rates, emergency surgery, and mortality rates for treatment of peptic ulcer bleed. Sung and colleagues¹²¹ reached a similar conclusion that hemoclips improved hemostasis compared with injection alone but were comparable to thermal coagulation. In a Cochrane review meta-analysis in 2007 that included 1763 patients, adding a second procedure reduced further bleeding rate from 18.8% to 10.4% (odds ratio [OR] 0.51), emergency surgery from 10.8% to 7.1% (OR 0.63), and mortality from 5% to 2.5% (OR 0.50). Subanalysis showed that risk decreased regardless of which modality was used as second therapy. Risk was reduced in all groups.¹²²

Role of Medical Therapy

Medical management targets reduction in morbidity, risk of rebleeding, need for transfusions, hospitalization duration, need for endoscopic or surgical intervention, and mortality. Testing and eradication for H. pylori is important in all patients with peptic ulcer bleeds. Several randomized controlled trials have shown that eradication of H. pylori in patients with bleeding peptic ulcers significantly decreases recurrent bleeding compared with nontreated patients.^131,132 In addition to endoscopic therapy, acid suppression therapy has been shown to benefit patients with bleeding peptic ulcers.¹³³ The role of gastric acid inhibition in stopping bleeding or preventing recurrent bleeding is related to sealant clot stability, which is favored at a higher gastric pH.^134,135 A pH greater than 6 is required for platelet aggregation, whereas clot lysis occurs at pH less than 6. IV and oral H₂ antagonists have previously been used to prevent recurrent bleeding in these patients. Two meta-analyses showed that H₂-receptor inhibitors provide no additional benefit in patients with bleeding duodenal ulcers, but statistically significant absolute risk reductions in rebleeding, surgery, and death were seen in patients with bleeding gastric ulcers.^133,136 Proton pump inhibitors (PPIs) have been found to be more effective than H₂-receptor antagonists in decreasing the rate of recurrent bleeding.^137,138

Lau and colleagues¹³⁹ reported results of a large double-blind randomized trial in which omeprazole was compared with placebo in patients receiving endoscopic therapy for bleeding ulcers (Table 12.8). After receiving endoscopic therapy (active bleeding, nonbleeding visible vessel, stigmata under adherent clot), patients were randomly assigned to IV omeprazole (8 mg/hr continuous infusion) versus placebo for 72 hours. Recurrent bleeding at 30 days, the primary endpoint of the study, occurred in 6.7% of the patients in the omeprazole group compared with 22.5% of the patients in the placebo group (relative risk reduction 70%, P < .001). Most cases of rebleeding occurred within 72 hours of endoscopic therapy. A Cochrane meta-analysis by Leontiadis and colleagues¹⁴⁶ on the role of PPI therapy in peptic ulcer bleeding included 24 trials with 4373 patients. Reductions in rebleeding, the need of surgery, and repeat endoscopic therapy were statistically significant with PPI use. Other meta-analyses reached similar conclusions.^147–150 Post hoc analysis of the Cochrane data by Leontiadis and colleagues¹⁵⁰ showed important differences between Asian and non-Asian patients. The effects of PPI therapy on rebleeding and need for surgery were more marked in the Asian trials, and PPI therapy was associated with significant reduction in 30-day mortality in Asian patients.

Lau and coworkers¹⁵¹ prospectively showed that IV bolus followed by infusion of PPI started before endoscopy decreased the need for endoscopic therapy, number of actively bleeding ulcers, and duration of hospitalization. A retrospective study from Keyvani and colleagues¹⁵² showed that PPI (oral and IV) started before endoscopy significantly reduced rebleeding, surgery, duration of hospital stay, and mortality in patients presenting with ulcer hemorrhage. In a meta-analysis, Laine and McQuaid¹⁵³ studied the role of PPI therapy as an adjunct to endoscopic therapy. Three groups were analyzed: (1) IV PPI (bolus followed by infusion) versus placebo, (2) IV PPI (bolus followed by infusion) versus H₂ blockers, and (3) oral or intermittent IV PPI versus placebo. Their conclusions were that PPI therapy after endoscopic therapy reduced rebleeding, need for surgery, need for urgent intervention, and mortality. The results were most consistent for bolus followed by continuous IV PPI for 72 hours.

Based on randomized clinical trials, the recommended dose of PPI is equivalent to omeprazole 80 mg IV bolus followed by 8 mg/hr infusion for 72 hours. The U.S. Food and Drug Administration (FDA) has not approved PPIs for the indication of upper GI bleed. In view of their known benefits and very good side-effect profile, starting PPI therapy seems to be a reasonable approach. Increasing data indicate that perhaps it is safe to switch the patient to oral PPI early in the course of management.^154,155 Somatostatin, octreotide, and tranexamic acid may have a theoretical advantage by their different mechanism of action of helping hemostasis, but there is no firm evidence to recommend them over PPI therapy for nonvariceal upper GI bleeding.

Second-Look Endoscopy

Routine repeat (second-look) endoscopy is done 24 hours after the initial procedure with the intent of repeat endoscopic treatment of residual high-risk lesions and has been advocated to be beneficial in some clinical situations. Few randomized trials have addressed the issue, and these trials show conflicting results. Messmann and colleagues¹⁵⁶ reported no improvements in outcomes when scheduled repeat endoscopy was compared with second-look endoscopy performed at recurrent bleeding. Villanueva and associates¹⁵⁷ noted a nonsignificant trend toward better outcomes in the group that received routine second-look endoscopy within 24 hours. Saeed and colleagues¹⁵⁸ adopted an approach of scheduled retreatment in high-risk patients based on a composite clinical and endoscopic score and showed significant benefit in the prevention of rebleeding.

Pooled data on second-look endoscopy showed that routine second-look endoscopy with retreatment significantly reduced the risk of recurrent bleeding compared with expectant management (absolute risk reduction 6.2%, P < .01, number needed to treat 16).¹⁵⁹ The risk of surgery and death were not significantly influenced by second-look endoscopy. Cost-effectiveness of repeat endoscopy has been verified in a hypothetical model,¹⁶⁰ but no prospective studies are available to date. Routine second-look endoscopy in all patients is not recommended. It may be justified in selected high-risk patients. Second-look endoscopy may be needed in patients with an incomplete initial endoscopy because of poor visualization or other technical difficulties.

Recurrent Bleeding

The size and the depth of ulcer influence the bleeding rate. Increased ulcer size (>1 cm) is associated with an increased risk of rebleeding and mortality. The failure rate of therapeutic endoscopy is much greater in ulcers larger than 2 cm.¹⁶¹ Ulcers located high on the lesser curvature of the stomach and on the posteroinferior wall of the duodenum also seem to have a higher rate of rebleeding.²⁴ Primary hemostasis is achieved in greater than 95% of patients with bleeding peptic ulcers and ulcers with a nonbleeding visible vessel. Rebleeding can occur in 10% to 20% of these patients with mortality of 4% to 10%. In a prospective study, Manguso and colleagues¹⁶² recruited patients admitted for nonvariceal upper GI bleeding with Forrest I lesions (active bleeding) at time of endoscopy. All patients received standard endoscopic therapy and IV PPI. The investigators reported a rebleeding rate of 12%, 2.8% of patients required surgery, and there was an overall mortality of 5.6% in a 24-day period.

Rebleeding occurs within 48 to 72 hours and is highly infrequent after 96 hours. In most patients, repeat endoscopy for diagnosis of the bleeding site and a renewed effort at endoscopic hemostasis is the next intervention. The number of repeated efforts at endoscopic therapy depends on various individualized factors. Repeat endoscopic intervention should include an evaluation for the selection of endoscope and therapy. The original presentation of the bleeding site, access to it, and the tissue response to the original therapy are important considerations. Angiography and surgery should be considered for patients with refractory recurrent bleeding and massive recurrent bleeding associated with hypotension, patients receiving transfusion of 4 or more units of blood for a resuscitative event, and patients receiving transfusion of 8 or more units of blood.

In a retrospective trial of patients with nonvariceal upper GI bleeding, rate of recurrent bleed was 20% after endoscopic therapy. The authors developed a score to predict rate of rebleeding in patients after endoscopic therapy. The score was based on six factors: failure to use a PPI after the procedure, endoscopically shown active bleeding, treatment with epinephrine monotherapy, peptic ulcer as a source of bleeding, IV heparin or low-molecular-weight heparin after procedure, and moderate or severe cirrhosis.¹⁶³

In a prospective randomized trial, Lau and coworkers¹⁶⁴ randomly assigned 92 patients with recurrent bleeding after endoscopic hemostasis; 48 patients were assigned to repeat endoscopic therapy, and 44 patients were assigned to emergency ulcer surgery. Of patients in the repeat endoscopic therapy group, 73% (35 of 48) had long-term control of bleeding by repeat endoscopic therapy; 27% (13 of 48) required surgery because of 11 endoscopic failures and 2 perforations secondary to thermal coagulation. Overall, a lower complication rate (14.6% vs. 36.4%) was seen in the repeat endoscopy group. This study suggests that repeat endoscopy is effective in reducing the need for surgical intervention and its associated complications.

During endoscopic retreatment with coagulation therapy, a large thermal probe (3.2-mm) is desirable. Repeat thermal therapy at the same location increases the risk of clinical perforation and should be used with caution in therapy of acute lesions, such as a Mallory-Weiss tear, stress ulcer, and Dieulafoy’s lesion in the gastric fundus, and in severely ill immunocompromised patients with poor tissue healing capability.¹⁶⁴ Injection therapies may be ineffective because of increased fibrosis, especially of chronic and penetrating ulcers. Injection therapy is reasonable to attempt during active spurting. If there is no immediate response, alternative therapy is needed. Refractory rebleeding in patients who are not candidates for surgical intervention or who have also failed angiographic intervention may be candidates for EUS-guided therapy by direct vascular injection of embolic material or glue. Fine needle application of tissue glues and even alcohol can be performed into specific types of lesions (e.g., gastrointestinal stromal tumors, pseudoaneurysms).¹⁶⁵

Complications of Endoscopic Therapy

The most feared complications of endoscopic hemostatic therapy are perforation and induction of uncontrollable bleeding. Both of these complications are relatively rare, however. One must be prepared for the possibility of inducing massive bleeding when a deeply penetrating ulcer is encountered with focally pulsating stigmata of a large visible vessel, especially when there is a fresh dense adherent clot attached to the stigmata. These are unique circumstances for which careful planning is required to ensure that blood products are readily available, endotracheal intubation is performed, and surgical backup has been readied.

Perforations are typically small (probe-sized), and efforts should be directed at closing the perforation with hemostatic clips. For larger perforations in the stomach, except posteriorly, and in the anterior duodenum, omental fat can be grasped and pulled into the defect, then clipped into place (omental patch). Laine and Peterson¹⁴ reported induction of uncontrollable bleeding requiring urgent surgery in 0.3% of cases and perforation in 0.5% of patients who underwent endoscopic therapy for peptic ulcer hemorrhage. In a meta-analysis by Cook and associates,⁷¹ endoscopic hemostasis induced bleeding requiring surgery in 0.39% and 0.4% of cases for contact thermal devices and injection therapy, and perforation rates of 0.7% and 0% were reported.

Failure of Endoscopic Therapy

Surgery for bleeding peptic ulcer has declined in the past 2 decades secondary to improvements in endoscopic hemostasis techniques and acid suppression therapy. However, active nonvariceal GI hemorrhage that cannot be controlled by endoscopic intervention and especially hemorrhage that overwhelms the ability to visualize any bleeding site requires an urgent operation. It is imperative to obtain surgical consultation in all patients presenting with the aforementioned clinical situation. Accurate prediction of patients likely to do poorly after endoscopic therapy is desirable. Wong and colleagues¹⁶⁶ analyzed risk factors associated with treatment failure with combined injection therapy and heater probe thermal coagulation. Hypotension, hemoglobin less than 10 g/dL, fresh blood in the stomach, ulcer with active bleeding, and ulcers larger than 2 cm were independent risk factors predicting poor outcome. In a prospective study by Chung and coworkers,¹⁶⁷ presence of spurting and ulcers larger than 2 cm were significantly related to failure of endoscopic therapy. Few data support elective surgery in these patients after their first endoscopic intervention, although it sounds appealing in theory because elective surgery carries a much lower mortality than emergency surgery. Maximal acid suppression therapy, close monitoring for signs of rebleeding, and surgical team on standby may be the best treatment strategy.

Prevention of Recurrent Bleed

Prevention of recurrent ulcer disease and bleeding is very important in patients with ulcer hemorrhage. Convincing data indicate that treatment with PPIs prevents ulcer rebleeding. Follow-up endoscopy is warranted to exclude malignancy in certain patients with gastric ulcer. The data linking persistent H. pylori with recurrent ulcer hemorrhage are compelling, making eradication of infection the best approach for these patients.^171–174 Most tests of active infection may exhibit increased false-negative rates in the context of acute bleeding.^175–178 The optimal diagnostic approach may include testing for H. pylori by antral biopsy at the time of endoscopy, with reconfirmation of negative results by repeat testing outside the acute setting of bleeding. Oral therapy can be started immediately or during follow-up in patients found to have H. pylori. Patients with peptic ulcer bleeding should be encouraged to discontinue use of NSAIDs. If stopping NSAID usage is impossible, therapy with misoprostol (200 µg four times daily) or omeprazole seems to be effective in prevention of gastroduodenal ulcers and erosions.^171,179 Switching to COX-2 inhibitors would still require concurrent prophylaxis therapy.

Dieulafoy’s Lesion

Dieulafoy’s lesion is an uncommon cause of acute nonvariceal GI bleeding, but it can be associated with massive, life-threatening hemorrhage. Originally described by Gallard in 1884 and designated “exulceratio simplex” by the French surgeon Dieulafoy 14 years later, Dieulafoy’s lesion consists of an abnormal, submucosal “caliber persistent artery” (1 to 3 mm diameter) that retains the large caliber and typically protrudes through a small mucosal defect without any surrounding ulceration.¹⁸⁰ The underlying mechanism is poorly understood but thought to be due to mechanical compression of overlying mucosa by the large artery resulting in small erosion with rupture of the vessel into the lumen.¹⁸¹ Dieulafoy’s lesion accounts for 1.2% to 1.9% of cases of acute nonvariceal upper GI bleeding,^182,183 but an incidence of 5.8% has been reported.¹⁸⁴ Although reported in all age groups, Dieulafoy’s lesion predominantly manifests in old men with multiple comorbid conditions.^181,184 Typically, a Dieulafoy lesion is located in the stomach usually within 6 cm of the gastroesophageal junction, the classic site, in 60% to 64% of cases and in the duodenum in another 14% to 18%.^184,185

Diagnosis

The endoscopic criteria for the diagnosis of Dieulafoy’s lesion are (1) active arterial spurting or micropulsatile streaming; (2) visualization of a protruding vessel with or without active bleeding; or (3) fresh, densely adherent clot with a narrow point of attachment, from a minute (<3 mm) mucosal defect or through normal surrounding mucosa (Fig. 12.5).¹⁸⁶ Diagnosis at the time of endoscopy can be difficult because there is no identifying lesion (ulcer) to indicate the source. The diagnosis rates for initial endoscopy range from 49% to 63%, and repeat endoscopy is often required.^182,187 In view of the intermittent nature of bleeding from Dieulafoy’s lesion, which can span years, the sensitivity of endoscopic diagnosis is increased through early endoscopy in patients with acute GI bleeding.

Fig. 12.5 Dieulafoy’s gastric lesions. A, Lesion with thin rim of surrounding ulceration. B, Pale protruding artery with surrounding blood. C, Pale centrally protruding artery with surrounding mucosal defect.

Dieulafoy’s lesion should be suspected in specific clinical situations. One clinical scenario involves a patient with repeated unexplained acute upper GI tract bleeding with repeatedly negative endoscopy. Another clinical scenario involves a patient who presents with a major acute bleed with the finding only of blood, fresh or altered in the stomach or duodenum. Repeat endoscopy timed as close as possible to a rebleeding event is crucial for the diagnosis. In a hospitalized patient in whom a gastric site of bleeding is suspected, a nasogastric tube can be useful to identify a rebleeding event promptly. EUS has been reported to identify Dieulafoy lesions successfully in this patient group and to guide therapy.

Mechanical Devices

Among mechanical devices, hemoclip application and endoscopic rubber band ligation have been successfully used to treat bleeding from Dieulafoy’s lesion. The rationale of using mechanical devices is the small size of the lesion with normal surrounding tissue, which can be compressed along with the caliber persistent artery.

Hemoclip Application

Hemoclip application hemostasis has been used for Dieulafoy’s lesion (Fig. 12.6) similar to other lesions. Yamaguchi and coworkers¹⁹² used hemoclip application as first-choice hemostatic treatment for bleeding from Dieulafoy’s lesion. Initial hemostasis was achieved in 94.1% of patients requiring 3.1 (mean, range 1 to 6) clips per patient. Rate of recurrent bleeding was 9.4%, which was successfully treated with repeat endoscopic therapy. None of the patients required surgery. Parra-Blanco and associates¹⁹⁰ used hemoclip application to control bleeding from Dieulafoy’s lesion in 69% of their patients. Initial success rate was 97% (17 of 18) with a mean of 2.7 clips per patient.

Fig. 12.6 Fundal Dieulafoy’s lesion treated by clipping.

In a randomized prospective trial, Chung and colleagues¹⁹³ compared mechanical methods (hemoclip and band ligation) with injection therapy with HSE (Table 12.9). Of 24 patients, 12 were assigned to mechanical therapy (9 hemoclip, 3 rubber band ligation), and 12 were assigned to the injection therapy group. Mechanical therapy was more effective than injection therapy in terms of initial hemostasis (91.7% vs. 75%), recurrent bleeding (8.3% vs. 33.3%, P < .05), and need for surgery (0% vs. 17%).

Table 12.9 Randomized Controlled Trials of Clipping for Dieulafoy Lesions

Endoscopic Rubber Band Ligation

Matsui and coworkers¹⁹⁶ compared endoscopic rubber band ligation with bipolar electrocoagulation in patients with acute GI hemorrhage exclusive of chronic gastroduodenal disease (Fig. 12.7). Endoscopic therapy was provided to 27 patients with Dieulafoy’s lesion. Success rate of hemostasis was 100% (13 of 13) for endoscopic band ligation versus 85.7% (12 of 14) for bipolar electrocoagulation. In a retrospective Veterans Affairs study, Mumtaz and colleagues¹⁹⁷ presented data of 23 patients with Dieulafoy’s lesion treated with endoscopic band ligation (14 of 23) and injection with or without thermal therapy. Initial hemostasis was achieved in all patients in both groups, and only one patient had early rebleeding (within 72 hours) in both groups. Nikolaidis and associates¹⁹⁸ retrospectively reviewed the results of endoscopic band ligation in 23 patients with Dieulafoy’s lesion. Initial hemostasis success rate was 96% (22 of 23). One patient with jejunal Dieulafoy’s lesion required surgery for rebleeding.

Fig. 12.7 Gastric Dieulafoy’s lesion (from Fig. 12.5C) treated by band ligation. Note very prominent translucent protruding artery.

Mallory-Weiss Tear

Mallory-Weiss tears are a common cause of nonvariceal upper GI bleeding and account for 3% to 10% of cases.^200,201 A Mallory-Weiss tear was found by endoscopy in 7% of patients in the ASGE survey.¹⁰ Mallory-Weiss tear is a mucosal laceration at the gastroesophageal junction or gastric cardia usually caused by retching or forceful vomiting. The underlying mechanism is thought to be rapid propulsion of gastric cardia into the thoracic cavity through the hiatus from retching or vomiting. If this is forceful enough, a longitudinal laceration can follow. Most patients with Mallory-Weiss tears are in their third to fifth decade, and there is a male preponderance in most case series.^200,201 Alcohol use is a common factor and is found in 40% to 70% of patients with Mallory-Weiss tear.^202–204 Many other factors have been associated with the development of Mallory-Weiss tears, including severe coughing, pregnancy, heavy lifting, straining, and colonic lavage using polyethylene glycol. Mallory-Weiss tears are well known, although uncommon complications of upper endoscopy.²⁰⁵

Future Directions for Endoscopic Therapy

Doppler ultrasound may provide objective findings in patients with ulcer hemorrhage. Doppler signal correlates with rebleeding suggesting that it may be useful to continue endoscopic treatment until underlying signal of blood flow stops.^213,214 Interobserver disagreement in interpretation and its availability preclude wide use of Doppler ultrasound at present. EUS-guided therapy has been used in sporadic cases to control refractory bleeding.¹⁶⁵ EUS-guided therapy has the advantage because therapy is delivered under ultrasound guidance, and direct visualization may not be needed. Hemoclips are the only mechanical device available for widespread clinical use, but it is expected that development of natural orifice transendoscopic surgery will offer additional endoscopic suturing devices that eventually will be applicable to endoscopic therapy of nonvariceal bleeding.²¹⁵