Gastrointestinal Intervention 2012; 1(1): 53-57
Published online December 30, 2012 https://doi.org/10.1016/j.gii.2012.09.002
Copyright © International Journal of Gastrointestinal Intervention.
James Y.W. Lau
Endoscopy Centre, Prince of Wales Hospital, The Chinese University of Hong Kong, Hong Kong Special Administrative Region, China
Correspondence to:*Department of Surgery, 4/F, Clinical Sciences Building, Prince of Wales Hospital, Shatin, Hong Kong Special Administrative Region, China.,
The management of patients with acute upper gastrointestinal bleeding should consist of early assessment and volume resuscitation. Endoscopy should be performed in all patients within 24 hours of their presentation. The Glasgow Blatchford score requires validation in different centers. It is accurate in identifying those at low risk of requiring intervention. The risk score is less specific in identifying those who require urgent endoscopic intervention. During endoscopy, the presence of active bleeding and a non-bleeding visible vessel mandate endoscopic hemostatic treatment. There is also evidence that clots overlying ulcers should be unveiled and hemostatic treatment offered to underlying stigmata. Injection therapy using diluted epinephrine alone is considered inadequate. A second treatment should be added to induce thrombosis of the bleeding artery. The use of thermo-coagulation with a thermal device or hemo-clips alone or after pre-injection with epinephrine is equally efficacious. Second look endoscopy should be performed in selected high risk or re-bleeding patients. Proton pump inhibitor (PPI) should administer as an adjunctive therapy. The dose of PPI use continues to be controversial. Angiographic embolization compares favorably to surgery as a rescue therapy where endoscopic therapy fails.
Keywords: Hematemesis, Melena, Peptic ulcer hemorrhage
Acute upper gastrointestinal bleeding (AUGIB) is a common medical emergency. In the UK, its incidence is between 80 and 110 per 100,000 in the population. Two national audits were conducted in the UK in 1993 and 2007 respectively.1,2 In over a decade, the crude overall mortality from AUGIB improved from 14 to 10%. The age structure has not changed significantly between the audits. In the second audit, 63 percent of patients were older than 60 years. Endoscopic therapy was performed in 24% (1172 of 4942) of patients. The rate of surgery has diminished from 6.7 to 1.9%. In the few who required surgery to stop bleeding, the mortality was 30%. The present review summarizes the current literature on the management of patients with AUGIB and specifically those with nonvariceal causes, the commonest being bleeding from a peptic ulcer, and makes suggestions for future research.
Patients who present with overt signs of upper gastrointestinal bleeding (melena or hematemesis with or without hypotension) should be assessed early. In many, bleeding has stopped by the time of presentation. The few with signs of ongoing bleeding should be volume resuscitated. Blood should be administered to unstable exsanguinating patients. In stable patients, clinicians should consider transfusing when hemoglobin level drops below 7 g/dl and aim for a level of around 8 g/dl. In patients with acute or chronic cardiac diseases, a hemoglobin level of around 10 g/dl should be the aim. There was evidence from the recent national UK audit that red cell transfusion was associated with increased risk of further bleeding across all risk categories as stratified by the Rockall score. In patients with a score of 6 to 8, further bleeding occurred in only 4% (12/288) patients without transfusion compared to 32% (292/917) patients with red cell transfusion. Policy in blood transfusion is now a subject of investigation in a large clinical trial in the UK. The TriGGer trial led by a research group in Oxford is a multicenter randomized controlled trial that compares liberal and restrictive transfusion policies.3
In patients actively bleeding and unstable on warfarin, an international normalized ratio (INR) <1.5 should be achieved quickly with the treatment of fresh frozen plasma or prothrombin complex. Low molecular weight heparin or heparin should be discontinued and protamine should be administered to those with a supra-therapeutic activated partial thromboplastin time and with signs of ongoing bleeding. This should not delay endoscopy with the objective to stop bleeding. The decision to reintroduce anticoagulation should be multidisciplinary, tailoring to individual needs and balancing risk of further bleeding and withholding anticoagulation therapy.
Although it is common sense to believe that aggressive resuscitation improves outcome, the only evidence in the literature comes from a comparative study using a historical cohort by Baradarian et al, which showed that aggressive resuscitation (shorter time to stable hemodynamics and to correct hematocrit >28 and INR <1.8) led to reduced death (1/36 vs. 4/36) and myocardial infarction (2/36 vs. 5/36).4
It is recommended that patients should be risk stratified on their presentations by the use of prognostic scores. The Rockall score is a composite score that combines pre-endoscopy information as well as findings at endoscopy.5 The score was derived from the first national UK audit to predict mortality from AUGIB. Blatchford and colleagues from Glasgow subsequently published a clinical score, based entirely on pre-endoscopy parameters.6 The Glasgow Blatchford Score (GBS) was used to predict the need for intervention, loosely defined by the need for transfusion or endoscopic therapy. GBS has been shown to be accurate in identifying low risk patients (GBS<1) for early discharge. Its role in triaging patients for be urgent out-of-hours endoscopy is less certain. We validated the use of both Rockall and GBS in our local population with upper gastrointestinal bleeding with the endpoint of predicting the need for endoscopic therapy.7 In our setting, the pre-endoscopic Rockall score was unable to predict such a need accurately. Although a high GBS (>1) was sensitive (100%) in identifying all patients who require endoscopic therapy, the specificity was only 6.7%. Many out-of-hours endoscopies would be performed if one used GBS = 1 as a cutoff point. While the GBS is a useful adjunct in deciding whether to offer urgent endoscopy, we rely on the clinical signs of ongoing bleeding (fresh hematemesis, shock with melena).
It is generally believed that early endoscopy improves clinical outcomes. This may well be true for individual patients with active bleeding, but the strength of evidence in the literature to support early endoscopy is not strong. Evidence in favor of early endoscopy (usually defined by endoscopy within 24 hours) comes only from cohort studies. Cooper et al reviewed hospital records from of 909 consecutive patients from 13 hospitals and found that early endoscopy was associated with reductions in recurrent bleeding, surgery (odds ratio of 0.21) and hospitalization (−31%) in high-risk patients.8 There have been three prospective randomized controlled studies, each consisting of small number of patients.9–11 The risk description in these studies was poor. Obviously even with a pooled analysis, these studies were not sufficiently powered to detect differences in recurrent bleeding, let alone mortality. It is clear that across all risk categories, early endoscopy of patients reduces hospitalization. Those at low risk can be considered for early discharge. A clinical trial is ongoing at our hospital that randomizes high-risk patients, defined as those with a GBS of >11 and with stable hemodynamics to endoscopy within 6 hours of admission or the next morning.
We conducted a large double blind placebo controlled randomized study in patients who presented with overt signs of AUGIB.12 Patients who could initially be stabilized were randomized to receive a high dose proton pump inhibitor (PPI) infusion or an equivalent placebo until endoscopy the next morning. The proportion of bleeding peptic ulcers was 60%. We showed a reduction in the need for endoscopic therapy in those who received a PPI infusion (19.1% vs. 28.4%) for a mean of around 14 hours. There were fewer actively bleeding peptic ulcers in those given PPI (12 of 187 vs. 28 of 190) coupled with more clean based ulcers (120 of 180 vs. 90 of 190). PPI use led to fewer high risk stigmata of bleeding. There was no difference in other clinical outcomes. PPI use prior to endoscopy should, however, not be misconstrued as a substitution for urgent endoscopy in those who need it. In stable patients waiting for endoscopy, PPI should be started early.
In an early meta-analysis of 30 randomized controlled trials, Cook et al provided convincing evidence that endoscopic therapy reduced rates of further bleeding, surgery, and mortality in bleeding peptic ulcers that were actively bleeding or exhibited a nonbleeding visible vessel (A protuberant discoloration or a sentinel clot).13 It is somewhat controversial whether to elevate a clot overlying an ulcer base. Endoscopists differ in their vigor in irrigating clots. In the literature, there are several definitions of an adherent clot. Some endoscopists would irrigate an ulcer with a thermal device for 5 minutes before declaring a clot ‘adherent’. Others would go as far as guillotining or ‘cheese-wiring’ the clot using a mini-snare. There have been two randomized controlled studies that compared endoscopic to medical therapy alone in ulcers with adherent clots. Both found a lower rebleeding rate (1/21 vs. 12/35 and 0/15 vs. 6/17) with endoscopic therapy.14,15 In a pooled analysis, Kahi et al summarized controlled trials on the subject and concluded that endoscopic therapy after clot elevation would reduce rate of further bleeding without impact on other clinical outcomes.16 Endoscopists should, however, be wary that a large subserosal artery may underlie a deep ulcer at high risk positions such as the bulbar duodenum and angular notch of the stomach. Clot elevation may provoke torrential bleeding. It may be prudent to recourse to alternative therapy such as angiographic embolization.
It is now clear that endoscopic epinephrine injection should not be used as a sole treatment. Calvet et al summarized 16 randomized controlled trials of 1673 patients that compared epinephrine injection alone to epinephrine injection followed by the addition of a second treatment modality.17 The latter reduced the rate of further bleeding from 18.4% to 10.6%, surgery from 11.3% to 7.6%, and, importantly, mortality from 5.1% to 2.6%. In another meta-analysis of 20 randomized trials that compared mono- to dual therapy, injection alone was shown to be inferior to dual therapy.18 When mechanical or thermal therapy was compared to dual therapy of injection plus either a mechanical or thermal treatment, the outcome differences were not statistically different. The two meta-analyses suggest that in the treatment of an actively bleeding vessel, if we use preinjection of diluted epinephrine to volume tamponade bleeding, a second treatment must follow to induce thrombosis of the bleeding vessel. If a vessel can be clearly seen without injection, one can either apply a thermal probe or a mechanical device such as a hemoclip. Often we use different and often several treatment modalities if one should fail. The choice is based on practical considerations. The application of clips can be difficult tangentially or with the scope in retroflexion in the difficult position of posterior bulbar duodenum and high lesser curvature. We summarized clinical trials that compared the use of hemoclips to thermal devices and found the rate of definitive hemostasis with either modality identical (pooled odds ratio of 1). Both are equally efficacious as hemostatic treatment.19
It has been a common practice in certain parts of the world such as Europe to perform second-look endoscopy the next morning in patients who receive endoscopic hemostatic treatment. El Ouali et al summarized results from four published articles and four abstracts consisting of 938 patients on randomized controlled study over the use of routine second-look endoscopy.20 Routine second-look endoscopy significantly reduced further bleeding (46/ 472 vs. 77/466,
Gastric neutrality is thought to be critical for clot stability. To maintain gastric pH consistently above 7, a high dose PPI infusion is required. Placebo controlled studies have confirmed that acid suppression can reduce rate of recurrent bleeding. The PUB study was a placebo-controlled multicenter international study that randomized 767 patients to receive either a high dose esomeprazole infusion or its placebo after endoscopic hemostasis to their bleeding peptic ulcers.22 The use of PPI led to reduction in recurrent bleeding in 72 hours (5.9% vs. 10.3%), further endoscopic treatment (6.4% vs. 11.6%) and surgery (2.7% vs. 5.4%). The study reaffirms findings of an earlier Cochrane systematic review that PPI as an adjunct to endoscopic treatment reduces further bleeding and improves outcomes.23
The optimal dose of PPI remains controversial. An Italian multicenter study that compared head to head high-to low-dose PPI found a similar rebleeding rate (11.8% and 8.1%) with either regimen.24 A meta-analysis of clinical trials that compared a high dose PPI infusion to a low dose (defined by 80 mg daily or below) given either intravenously or orally suggested that a low dose regime would be as efficacious.25 The pooled analysis has to be interpreted with caution. Several of the trials included consisted of patients with low risk stigmata of bleeding and even clean-base ulcers. The inclusion of such ulcers would dilute the potential benefit of a high dose regime. A fair comparison between low and high dose PPI regimes should enroll consecutive patient including those at high risk of rebleeding and death. Ideally the trial should be of a double dummy double blind design to test equivalence and with narrow margins of accepted difference. More trials have emerged to indicate a high dose is better than a low dose.26 No doubt the controversy will continue. In the meantime, we have continued to use a high dose regime particularly in high-risk patients.
Patients with clean-base ulcers or ulcers with flat pigments are often discharged home following endoscopy. A low GBS (≤1) is thought to be accurate to identify those at low risk for early discharge. Cooper and colleagues27 recently examined outcomes of 9123 episodes of bleeding in patients with upper gastrointestinal bleeding managed as in- or out-patients (38.4%) in a medicare database. The overall 30 day mortality rate was 8% in the inpatient group and 6.3% in the outpatient group. The high mortality associated with an outpatient care is alarming. We retrospectively analyzed causes of death in a cohort of 9375 patients with peptic ulcer bleeding. The mortality rate was 6.2%, 79% of which were nonbleeding related.28 This underlies the fact that clinicians are often faced with old patients with significant comorbid illnesses. An episode of bleeding may herald another medical event such as a cardiovascular accident. Patients are often required to be hospitalized for optimization of their medical conditions.
There have been improvements in interventional radiology mostly in microcatheter techniques and new embolic agents. Angiographic embolization should be the next treatment when endoscopy therapy fails. Six retrospective studies compared outcomes following transarterial embolization to surgery in patients with uncontrolled bleeding at endoscopy and those who rebled after initial control.29–34 Mortality following angiographic embolization and surgery varied from 3–25.8% and 14–30%, respectively. In all studies, morbidities, mostly cardiovascular, were higher in those who received surgery. Complications directly related to angiographic embolization are uncommon (from 0–10% in most series). These include duodenal ischemia and secondary stenosis, gastric, hepatic or splenic infarction, groin complications, and renal impairment. Our group is conducting a prospective randomized controlled trial comparing the two modalities as rescue procedures in patients with uncontrolled bleeding during endoscopy. In addition, we advocate the use of angiographic embolization to ulcers predicted at high risk of rebleed after initial endoscopic hemostasis. Elmunzer et al summarized the literature and found that some of the predictors for failures of endoscopic treatment include large ulcer size, posterior bulbar, and lesser gastric curvature ulcers in addition to shock and presence of comorbid illnesses.35 The high-risk ulcers located at bulbar duodenum and lesser curve can erode into a subserosal artery from the gastro-duodenal or left gastric artery complex. We believe that a prophylactic approach using angiographic embolization may reduce significantly risk of recurrent bleeding and impact upon patient outcomes. We use a sandwich technique in excluding the bleeding point from an artery. In the case of the gastro-duodenal artery, coils are deployed into the confluence between superior pancreatico-duodenal artery and right gastro-epiploic artery. Gelfoam particles are then used to pack the mid portion of the gastro-duodenal artery before coils are again left in the proximal part of the artery. The technique blocks off back bleeding from collaterals and reduces the risk of duodenal ischemia as the use of coils is generally quite safe.
All patients with bleeding peptic ulcers should be tested for
There may be false-negative results in
Among patients with an ulcer bleed, treatment either with a COX-2 inhibitor or a NSAID plus a PPI is associated with a significant risk of recurrent ulcer bleed. Our group studied patients who needed NSAIDs for arthritis and presented with ulcer bleeding.37 They were tested
Aspirin should be restarted early in patients with a high vascular thrombotic risk. We randomized 158 patients with aspirin-induced ulcer bleeding to receive aspirin or its placebo after endoscopic hemostasis and a PPI infusion for 3 days.39 Reintroduction of aspirin led a two-fold increase in recurrent bleeding. The all-cause mortality increased significantly when aspirin was withdrawn (1.6% vs. 14.1%). The relative risk for mortality was 1.7-fold greater in the no aspirin group. Cardiovascular thrombotic complications often occur between 6 and 10 days after withholding aspirin. Inhibited platelets circulate in the blood for about 10 days. In patients whose ulcers harbor low risk stigmata, aspirin should be started soon after endoscopy. In almost all cases, aspirin can be started after 3 days.
The CAPRIE study published in 1996 prompted many clinicians to switch patients from aspirin to clopidogrel.40 This was a randomized study designed to compare aspirin 325 mg to clopidogrel 75 mg taken daily in the prophylaxis against composite vascular event. A relative risk reduction of 8.7% was observed associated with the use of clopidogrel (5.3 vs. 5.8% with mean follow-up of 1.9 years in 19,185 patients). Two independent randomized controlled studies from Hong Kong compared the use of aspirin plus a PPI to clopidogrel alone in patients who had a prior bleed and were
Patients with drug eluting coronary stents require dual anti-platelet therapy. Stent thrombosis is often fatal. Antiplatelet therapy should be resumed as early as possible after an episode of bleed especially those with low risk stigmata and in whom stents were placed recently (within 6 months). Patients who continue to require dual antiplatelet therapy and at risk of further bleeding should be prescribed a PPI. There has been concern of drug interaction between clopidogrel (a thienopyridine derivative that blocks ADP receptors on platelets) and PPI. Both are metabolized by CYP2C19 isoenzymes of the cytochrome P450 system in the liver. The Clopidogrel and the Optimization of Gastrointestinal Events Trial (COGENT) provided assurance that co-therapy with PPI may be a better strategy.43 The trial randomized 3873 patients to receive dual therapy with and without a PPI and found that the gastrointestinal events were fewer with PPI (1.1% vs. 2.9% at 180 days). The cardiovascular event rate was 4.9% with PPI and 5.7% with placebo. The prescription of a PPI less dependent than CYP2C19 is probably desirable.
It is often said that despite of all advances, mortality from AUGIV has remained unchanged. This is untrue as evident from the improved mortality figure between the two national UK Audits. Further improvements may come in several areas. First, we are seeing better strategies secondary prophylaxis against recurrent bleeding. Second, patients should have easier access to care especially expertise in endoscopic therapy through better organization of acute care. Third, recurrent bleeding and failed endoscopic therapy predict a high mortality. This subgroup should be identified early. Targeted care and early angiographic embolization to the bleeding artery may be the way forward. The contemporary care of patients with AUGIB should be multidisciplinary involving not only endoscopists, interventional radiologists (very occasionally surgeons) but also transfusion and ICU specialists in managing coagulopathy and organ support. There will be new threats from drugs such as Dabigatran and Rivaroxaban.
Endoscopic and PPI therapies are cornerstones to management of AUGIB. The current debated issues are; whether early endoscopy improves clinical outcomes and the paradoxical effect of blood transfusion to recurrent bleeding. The management of patients on antithrombotic drugs is complex. We lack clinical evidence and management guidelines in these patients. Clinical research should target patients predicted to fail endoscopic treatment.
© The Society of Gastrointestinal Intervention. Powered by INFOrang Co., Ltd.