Nonvariceal Upper Gastrointestinal Bleeding

Introduction

The annual incidence of upper gastrointestinal bleeding (UGIB) is 48 to 160 events per 100,000 adults in the United States, where it is the cause of approximately 300,000 hospital admissions per year. In Europe, the annual incidence of UGIB in the general population ranges from 19.4 to 57.0 events per 100,000 individuals. There is no formal explanation for the breadth of the range between countries, although differences in health care systems and case recording capacities may be substantial. Additional contributing factors may include alcohol intake and H. pylori prevalence.

In both North America and Europe, in a general practice setting, approximately 80%–90% of acute UGIB episodes have a nonvariceal etiology with peptic ulcers and gastroduodenal erosions accounting for the majority of lesions. UGIB-related mortality rates have decreased in the past two decades, but still range from 2% to 15%. Two studies conducted in the United Kingdom highlighted that, despite a notable decrease in mortality in patients with UGIB from 1993 to 2007, mortality from peptic ulcer bleeding is still 10%–13%.

This chapter discusses the various management principles of medical therapy in nonvariceal UGIB (NVUGIB) put forth by guidelines from the multidisciplinary international consensus group in 2010, the American College of Gastroenterology in 2012, and the European Society of Gastrointestinal Endoscopy in 2015. When applicable, more recent evidence is included with a focus on randomized controlled trials (RCTs) and metaanalyses. Each section of this chapter presents the most current evidence and provides recommendations for practice. This chapter focuses on providing an evidence-based approach to initial resuscitation, the role of blood transfusion, preendoscopic, endoscopic, and postendoscopic management for patients presenting with NVUGIB.

Clinical Presentation

The presentation of GI bleeding is dependent on the site, volume, and rate of bleeding. The most common clinical presentation of UGIB, defined as a source proximal to the ligament of Treitz, includes either hematemesis or melena. Occasionally, the emesis can appear to resemble coffee-grounds, which can occur when older blood in the stomach is reduced by stomach acid. The appearance of melena (black, tarry, and foul-smelling stool) is caused by the degradation of blood as it passes through the small bowel and colon. Bleeding that originates from the small bowel or proximal colon can also present with melena. Importantly, a massive UGIB can manifest as hematochezia (bright red blood per rectum) since the rapid passage of blood through the GI tract does not have time to be degraded. Presentation with hematochezia accounts for approximately 10% of patients with NVUGIB and is more common in patients presenting with hemodynamic instability.

Initial Evaluation

The immediate priority in management is to secure the patient's airway, and tend to breathing and circulation. The patient should be placed in a monitored setting, at which point large-bore venous access should be established. Resuscitation should be initiated for patients with NVUGIB prior to any other procedure, and should include stabilization of the blood pressure with appropriate infusion of sufficient fluid volumes. In the intensive care unit (ICU), saline or Ringer acetate is preferred to hydroxyethyl starch (HES), as HES has been shown to increase the need for renal-replacement therapy in ICU patients, and may increase severe bleeding.

The primary objectives of resuscitation are to restore blood volume and to maintain adequate tissue perfusion, in the hope of preventing hypovolemic shock and ultimately death. No data suggest that any particular type of colloid solution is safer or more effective in patients needing volume replacement. Certain patients will require resuscitation with blood products including red blood cells, platelets, and, rarely, clotting factors (e.g., fresh frozen plasma).

Laboratory Data

Certain laboratory tests are required at initial presentation to guide resuscitation and to risk-stratify patients based on established scoring systems. Initial blood work should include: hemoglobin level, platelet count, blood urea nitrogen, creatinine, prothrombin time, and partial thromboplastin time. A type and screen or cross-match is indicated if transfusion is being considered. Other investigations including liver enzymes and liver function testing, cardiac enzyme testing, and an electrocardiogram are often performed to exclude potential confounding diagnoses or complicating conditions. Additional investigations can be guided by individual patient characteristics noted on history or physical examination.

Role of Nasogastric Aspirate

The routine placement of a nasogastric tube (NGT) in patients presenting with NVUGIB remains controversial and is not recommended by current guidelines. One 2013 retrospective study of 166 patients was able to predict active bleeding at endoscopy by combining NGT aspirate results with blood pressure and heart rate parameters. Opinions vary as high-quality evidence is lacking to decide whether certain patients benefit from NGT aspiration and lavage to help aid in both diagnosing UGIB, as well as improving visualization during endoscopy. Studies have failed to show any improvement in clinical outcomes attributable to the insertion of an NGT. One study identified 520 patients with UGIB who had a documented nasogastric aspirate prior to endoscopy and found that a bloody aspirate was associated with high-risk lesions (odds ratio [OR] 4.82) although the negative predictive value was only 77.9%. Importantly, a clear nasogastric aspirate reduced the likelihood to 0.15. There are, however, certain patients in whom the source of GI bleeding is unclear and who may benefit from placement of an NGT to confirm an upper GI source. Current guidelines do not suggest routine use of NGT placement prior to endoscopy.

Risk Stratification

Several scoring systems have been created to risk-stratify patients presenting with NVUGIB. Patients who are deemed high-risk may benefit from earlier and more aggressive treatments, including admission to the ICU and more rapid endoscopic evaluation. Moreover, low-risk patients may be appropriate for outpatient management, which could be safe and more cost-effective. Each scoring system has strengths and limitations that we will briefly outline hereafter. For detailed descriptions of the components of the most studied scores, see Table 14.1 .

The Rockall score has been shown to accurately predict rebleeding and mortality rates for patients presenting with NVUGIB. This scoring system incorporates clinical and endoscopic information to provide a total score for each patient. Multiple studies have validated the Rockall score and suggest that patients with a low score have a lower likelihood of rebleeding and mortality. Specifically, a score of less than 3 predicts a rebleeding rate of approximately 5% and a mortality rate of 0%. The major limitation of the Rockall score is that endoscopy must be performed to calculate the total score, thus limiting its usefulness in preendoscopic risk stratification.

The Glasgow-Blatchford score (GBS) uses clinical data on presentation to determine whether a patient with NVUGIB will require a blood transfusion, endoscopic intervention, or surgery. Several prospective studies have evaluated whether certain patients with low GBS scores can be discharged safely from the emergency department without immediate endoscopic evaluation. One prospective, multicenter study found that patients presenting to the emergency department with a GBS score of 0 could be safely managed as outpatients because none of these patients returned to their index hospital with recurrent UGIB or death within 6 months. More recently (2012), a metaanalysis found that a GBS score of 2 or more was 98% sensitive at identifying patients who require urgent evaluation.

The AIMS65 score represents one of the most recently developed tools for risk-stratifying patients and may be superior to previous scoring systems with regards to predicting inpatient mortality. Using only 5 clinical variables available at presentation, it is more accurate at predicting mortality than the GBS score. The AIMS65 score, however, was less accurate at predicting the need for blood transfusion than GBS score, and similar in predicting in-hospital rebleeding.

Limitations of all three scores include disparate comparative study results, variations in performance based on study populations and (surprisingly) geographic regions, and the absence of true interventional trials assessing patient outcomes following the introduction of a score early on in patient management.

Red Blood Cell Transfusion

The decision whether to transfuse red blood cells is based on the presenting hemoglobin level, hemodynamic status, patient comorbidities such as cardiac disease, and symptoms of anemia. A large Cochrane review put into question the usefulness of red blood cell transfusions in UGIB, demonstrating no overall survival benefit.

A 2013 single-center RCT of 921 patients with acute UGIB found a restrictive transfusion strategy (transfusion threshold < 7 g/dL with target hemoglobin of 7–9 g/dL) significantly decreased 6-week mortality, length of stay, and transfusion-related adverse events compared with a liberal transfusion strategy (transfusion threshold < 9 g/dL with target hemoglobin of 9–11 g/dL). The overall mortality benefit conferred by the restrictive transfusion strategy appears to have been driven by results obtained in patients with Child-Pugh class A and B cirrhosis. The subgroup of patients with NVUGIB did not exhibit a significant decrease in their overall mortality with a restrictive transfusion policy, although there was a trend towards benefit with no suggestion of harm. Importantly, this study excluded all patients presenting with severe hemorrhagic shock, acute coronary syndrome, symptomatic peripheral vasculopathy, stroke, or transient ischemic attack in keeping with consensus guidelines, suggesting higher hemoglobin targets for some of these patients.

Current guidelines suggest that patients with hemoglobin levels of 7 g/dL or less should receive blood transfusions to reach a target hemoglobin level of 7–9 g/dL, provided that the individual has no coronary artery disease, evidence of tissue hypoperfusion or acute hemorrhage. In patients with acute coronary syndrome, UGIB is associated with a markedly increased mortality, and a higher hemoglobin target level, above 10 g/dL, may be required.

As part of a recent UK audit, despite 73% of patients with UGIB presenting with a hemoglobin level greater than 8 g/dL, approximately 43% received red blood cell transfusions. A recently published metaanalysis (2014) not restricted to NVUGIB suggests that a restrictive transfusion approach reduces health care associated infections. Additional randomized trials are needed to address the issue of hemoglobin transfusion thresholds in patients with NVUGIB, including patients with significant cardiac disease. One large multicenter randomized trial is currently being completed in the United Kingdom, which may help answer such uncertainties.

Reversal of Anticoagulation

Many patients who present with NVUGIB will be on anticoagulation for a variety of reasons, most commonly due to atrial fibrillation. Of these patients, many are on warfarin, which can be monitored using the international normalized ratio (INR). Several retrospective studies have shown no difference in rebleeding, mortality, length of stay, or transfusion requirements between patients with an INR between 1.3 and 2.7. One cohort study compared patients with UGIB not on anticoagulation to patients on warfarin that had their INR corrected to 2.5 or under using fresh frozen plasma. They found no differences in rebleeding, surgery, mortality, or complication rates between the two groups. As such, upper endoscopy (EGD) in patients with therapeutic INR levels on warfarin appears to be safe to allow for early endoscopy. Current guidelines suggest that endoscopy should not be delayed to correct anticoagulation in patients on warfarin who have a therapeutic INR. For patients with an INR above 3, rapid reversal can be considered with either prothrombin complex concentrate or fresh frozen plasma in addition to vitamin K.

Several new direct-acting oral anticoagulants are commonly prescribed to patients given their higher efficacy at preventing stroke and systemic thromboembolism in patients with atrial fibrillation. These direct-acting anticoagulants now account for approximately 62% of all prescriptions for nonvalvular atrial fibrillation within the United States. Some of these drugs may increase GI bleeding when compared to warfarin. One major concern with these direct-acting oral anticoagulants is that there is no reliable way to monitor their efficacy on routine blood work. Furthermore, only dabigatran (Pradaxa) currently has an approved reversal agent, known as idarucizumab (Praxbind), whereas reversal for the other agents await commercialization in most countries. At this time, there are no clear guidelines as to the precise management of NVUGIB patients taking these newer anticoagulants. There have been some reports of clotting factors such as fresh frozen plasma, activated factor VII, or prothrombin complex concentrate being used for partial reversal of anticoagulation. Experts have suggested that early consultation with a hematologist and/or cardiologist should be considered in all patients with suspected GI bleeding on any of the direct acting oral anticoagulants.

Timing of Endoscopy

The timing of endoscopic evaluation depends on patient factors (hemodynamic stability, airway protection) and hospital factors (monitored setting, availability of endoscopic expertise, dedicated nurses and equipment). Several studies have confirmed that emergency endoscopy (within 6–12 hours) for UGIB has no advantage in terms of rebleeding, surgical rates, or mortality when compared to routine endoscopy within 24 hours. In one large retrospective study of 934 patients, Lim et al (2011) found that early endoscopy (defined as endoscopy within 13 hours of presentation) was associated with a lower mortality rate for high-risk patients with GBS scores of 12or higher. Interestingly, there was no difference in rebleeding rate, transfusion requirements, or surgery. Moreover, for low-risk patients with GBS scores less than 12 there were no differences in any outcomes when comparing early endoscopy to endoscopy within 24 hours of presentation. There is ongoing controversy as to the existence of a “weekend effect” whereby mortality from UGIB may be higher after regular working hours compared with patients treated at other times. Some authors postulate this effect may be due to a patient selection bias (sickest presenting at any time, including after hours) or decreased resources including delays to endoscopy and other treatments.

Current guidelines suggest that endoscopy should be performed within 24 hours in all patients presenting with UGIB who are admitted to hospital because this practice has been shown to decrease length of stay, rebleeding rates, and need for surgery. Some experts believe that, based on the limited evidence, endoscopy should be considered within 12 hours in selected high-risk patients.

Preendoscopic Proton Pump Inhibitors

Despite theoretical pharmacological differences between the different proton pump inhibitors (PPIs), no data support the use of a particular intravenous PPI over another when treating patients with UGIB. In the rest of this chapter, PPI will be used as a generic term for all such agents.

Starting PPI treatment before endoscopy for UGIB remains a controversial practice. A metaanalysis that included 2,223 participants from six randomized, clinical trials found that preendoscopic PPI treatment reduces the proportion of patients identified as having high-risk lesions (active bleeding, non-bleeding visible vessel, and adherent clot) thus reducing the need for endoscopic therapy (i.e., down-staging of high-risk endoscopic lesions). Despite these advantages, there is no evidence that preendoscopic PPI treatment affects mortality, risk of rebleeding, or need for surgery. For this reason, current guidelines recommend initiating PPI therapy on presentation to hospital, although it should never delay optimal resuscitation. PPI therapy pre-endoscopy is especially cost-effective in certain scenarios: when endoscopy is delayed for more than 16 hours after admission, or if patients have a high likelihood of nonvariceal bleeding, especially in those with high-risk symptoms, such as hematemesis. Current guidelines advise against the use of histamine-2 receptor antagonists in acute ulcer bleeding.

No recommendations can be made regarding the optimal dose or optimal route of administration of PPIs administered preendoscopy. A reasonable strategy may be to adopt a high-dose intravenous bolus (e.g., 80 mg) followed by a continuous infusion (e.g., 8 mg/hour) regimen. This infusion can be continued until endoscopy, at which point reassessment is required depending on the appearance of the bleeding lesion and the need for therapeutic intervention (see section on postendoscopic PPI therapy). A recent budget impact analysis assessing PPI institutional costs both pre- and postendoscopy suggest minimal incremental costs when opting for high-dose PPI infusion ( Fig. 14.1 ).

Etiology of Nonvariceal Upper Gastrointestinal Bleeding

The most common causes of NVUGIB include: gastroduodenal ulcers, angiodysplastic lesions, Dieulafoy lesions, Mallory-Weiss tears (MWT), malignancy, esophagitis, and erosive gastritis among other causes. Endoscopy allows for both the diagnosis and treatment of these lesions. The relative frequency of the causes of UGIB from a recent audit of 5004 patients presenting with UGIB can be seen in Table 14.2 .

Indications for Endoscopic Therapy for Gastroduodenal Ulcers

The role of endoscopic therapy for the management of gastroduodenal ulcers is based on the Forrest classification, which categorizes ulcer into high and low risk in terms of rebleeding risk ( Table 14.3 ). Several studies have since validated the Forrest classification in terms of predicting risk of rebleeding and the need for endoscopic intervention. Overall, endoscopic therapy is required for patients with ulcers classified as IA, IB, or IIA, and is controversial for patients with an adherent clot (see section titled Nonbleeding Adherent Clot ). For ulcers with IIC and III classification ( Fig. 14.2 ), endoscopic therapy is not warranted, as rebleeding risks are low ( Table 14.4 ; Fig. 14.3 ).

TABLE 14.3

Risk of Rebleeding Peptic Ulcer Disease Despite Medical Management Alone Based on Forrest Classification on Initial Endoscopic Evaluation

Endoscopic Stigmata (Forrest)	Frequency	Rebleeding on Medical Management
Spurting vessel (IA)	10%	90%
Oozing vessel (IB)	10%	10–20%
Nonbleeding visible vessel (IIA)	25%	50%
Adherent clot (IIB)	10%	25–30%
Flat pigmented ulcer base (IIC)	10%	7–10%
Clean-based ulcer (III)	35%	3–5%

TABLE 14.4

Forrest Classification for Endoscopic Appearance of Gastroduodenal Ulcers

Forrest Classification	Endoscopic Appearance	Endoscopic Therapy
IA	Spurting vessel	Therapy required
IB	Oozing vessel	Therapy required
IIA	Nonbleeding visible vessel	Therapy required
IIB	Adherent clot	Consider therapy
IIC	Flat pigmented ulcer base	No therapy required
III	Clean-based ulcer	No therapy required

Nonbleeding Adherent Clot

The endoscopic management of an adherent clot overlying an ulcer base remains controversial in the literature ( Fig. 14.4 ). Consideration must be given to endoscopic removal of the clot and treatment of the underlying lesion if it satisfies criteria for high-risk stigmata (e.g., Forrest classification IA, IB, or IIA). Vigorous irrigation of the clot using a water pump has been shown to expose the underlying stigmata in 26%–43% of cases, of which 70% were high risk stigmata requiring endoscopic therapy. The rebleeding risk for clots that remain despite irrigation has been estimated between 0%–35% in several studies. One metaanalysis of 6 RCTs comprising 240 patients with adherent clots found that although endoscopic therapy was associated with a lower rebleeding risk than medical therapy alone (8.2% vs. 24.7%, p = 0.01), other endpoints, including length of hospital stay (6.8 vs. 5.6 days; p = 0.27), transfusion requirements (3.0 vs. 2.8 units; p = 0.75), and mortality (9.8% vs. 7%, p = 0.54), were no different between the two groups. This metaanalysis did have several limitations including the vastly different patient populations, and statistical tests used to assess the impact of heterogeneity.

Endoscopic therapy for adherent clots consists of preinjection with epinephrine followed by removing the adherent clot using a cold snare. Once the clot is removed, combination therapy can be applied to the underlying stigmata as indicated. One 2015 observational study suggested that endoscopic therapy was preferred over conservative therapy alone for adherent clots in terms of mortality but did not demonstrate lower rebleeding rates. Future multicenter randomized controlled studies are required to better identify patients who may benefit from endoscopic intervention compared to PPI therapy alone.

Modalities for Endoscopic Therapy

The most commonly used endoscopic techniques for management of a bleeding ulcer include: injection therapy, thermal coaptive therapy, endoscopic clipping, and hemostatic powders. Each method has its strengths and weaknesses and often more than one technique can be used for a given case. Each modality will be discussed separately, including the role of combination therapy.

Injection Therapy

Epinephrine is the most established injection agent used for peptic ulcer injection therapy, although studies have been done using normal saline along with a sclerosant agent. The mechanism of action is thought to be local tamponade and vasoconstriction ( Fig. 14.5 ). Epinephrine diluted to 1 : 10,000 in normal saline is found to be most effective and safest. A large RCT of 165 patients with actively bleeding ulcers or visible vessels was performed to determine the ideal amount of epinephrine injection (1 : 10,000). This RCT, which was largely representative of the existing literature, found that large volume injection (13–20 mL) compared to small volume injection (5–10 mL) was associated with less rebleeding (15.4% vs. 30.8%, p = 0.037). A large metaanalysis of 1673 patients revealed that epinephrine injection alone is inferior to either clips or thermal therapy. An additional metaanalysis confirmed these findings. One of the roles of epinephrine injection is to temporize and control bleeding to achieve better visualization and so apply more definitive therapy (e.g., thermal therapy or clipping) to the lesion. Overall, injection therapy is a useful technique to incorporate into the management of ulcer bleeding but should not be used as monotherapy.

You're Reading a Preview

Become a Clinical Tree membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here