Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

7.1 Introduction

Upper gastrointestinal haemorrhage (UGIH) is a frequent cause of acute hospital admission, and in the United Kingdom, it accounts for 70,000 admissions per year with the majority of cases being non-variceal [1] (Table 7.1). Gastroduodenal (peptic) ulcers are the most common cause and account for well over 50 % of admissions. This is despite the ready availability of proton pump inhibitors and the recognition of the role of Helicobacter pylori and nonsteroidal anti-inflammatory drugs (NSAIDs) in their genesis.

Table 7.1 Aetiology of upper GI haemorrhage
Full size table
Fig. 7.1
figure 1

Endoscopic view of an ulcerated GIST. The patient presented with haematemesis and melaena

Full size image
Fig. 7.2
figure 2

Ulcer in a segment of columnar-lined oesophagus (Barrett’s ulcer); a rare cause of an acute upper GI haemorrhage

Full size image

Although there have been significant advances in endoscopy and interventional radiology (IR), UGIH remains a significant cause of morbidity and mortality. In fact, the 30-day mortality seems unchanged at a level of around 11 % as the patients tend to be older with more serious co-morbidities [2, 3]. Management has evolved over the last few decades with fewer cases requiring surgery, but therapy will depend on the place of treatment as smaller hospitals may not have the equipment or expertise to use the most modern techniques. In addition, in some health systems, gastroenterologists manage the patient, and surgical involvement is limited to severe cases requiring operation that have failed endoscopy and IR.

UGIH is defined as bleeding proximal to the ligament of Treitz and may present with melaena or haematemesis. Milder forms may present with anaemia and non-specific symptoms. The majority of cases are due to peptic ulceration, but the initial management and general response to non-surgical treatments are similar irrespective of the aetiology. This does not include variceal haemorrhage due to portal hypertension as this requires a different approach from the outset although patients with varices can also bleed from other lesions.

There still remain a small number of patients who require emergency surgery, and for the emergency general surgeon, this can be quite a challenging problem due to a lack of experience with elective gastroduodenal surgery.

7.2 Initial Management

The classic clinical presentation of a significant UGIH is a patient with haematemesis and melaena. A full medical history should be taken including a detailed medication history. Ulcerogenic drugs (NSAIDS, aspirin, steroids), antiplatelet drugs and anticoagulants (warfarin, rivaroxaban) and B-blockers are especially important to note. Discussion with haematologist should be undertaken in those who are on the newer antiplatelet or anticoagulant medications or when massive transfusion is likely.

All standard resuscitation measures should be instituted including transfusion. A restrictive transfusion policy aiming for a haemoglobin of 70–80 g/L is suggested in haemodynamically stable patients [4]. Gastric acid suppression should be achieved by intravenous proton pump inhibitor (PPI) bolus followed by infusion (e.g. omeprazole 80 mg stat then 8 mg per hour for 72 h). There appears no place for routine use of nasogastric tubes or lavage, antifibrinolytics (e.g. tranexamic acid) or octreotide. There is some evidence that erythromycin acting as a prokinetic (motilin agonist) improves visualisation at emergency endoscopy [4]. CT (computed tomography) angiography has a role in the subgroup of patients who are suspected of having an unusual cause of UGIH, e.g. tumour or aortoenteric fistula.

7.3 Endoscopy

Endoscopy is the critical step in managing patients with UGIH. It allows diagnosis, gives prognostic information and can control bleeding. It has been shown in randomised studies of peptic ulcer to lead to a reduction in blood transfusion, shortened intensive care unit and hospital stay, decreased need for surgery and lower mortality rate [5]. Endoscopic haemostasis is now accepted as the first-line treatment in patients with active bleeding. Several controlled trials have shown that endoscopic therapy using a variety of combined techniques significantly reduces the need for blood transfusion and emergency surgical intervention [6]. While most studies relate to peptic ulceration, endoscopic therapy has been shown to be effective in treatment of other causes, in particular, Dieulafoy or Mallory-Weiss lesions.

The Forrest classification of endoscopic appearance of peptic ulcers is used to give prognostic information: F1a spurting haemorrhage, F1b oozing haemorrhage, F11a non-bleeding visible vessel, F11b adherent clot, F11c flat pigmented spot and F 111 clean base [7]. Endoscopic therapy should be used when there is active bleeding or a visible vessel. For F11b lesions, there is no consensus as studies have conflicting results, and the decision whether to remove the clot to allow therapy should be made on a case-by-case basis. The therapy will depend on the experience of the endoscopist and the equipment available and include injection of epinephrine, heater-probe coagulation, bipolar electrode coagulation, laser coagulation, argon plasma coagulation, endoscopic clips and banding devices [8, 9].

Patients presenting with an UGIH who stabilise quickly with minimal resuscitation should have endoscopy within 24 h [10]. Debate continues on the relative merits of emergency endoscopy in these patients as there is increased risk of the procedure and it may be suboptimal due to blood and clots making therapy difficult [11]. Patients who are well and are found to have low-risk ulcers probably do not need inpatient monitoring. The use of risk scoring systems may help to identify the low-risk patients suitable for outpatient management; however, a recent study from Denmark found that none of the popular risk scoring systems examined were suitable for predicting risk of rebleeding or 30-day mortality, possibly due to inter-country variation in patient characteristics [2].

Patients with severe, life-threatening haemorrhage not responsive to resuscitative efforts need to be transferred to the operating room for emergency endoscopy. These patients should have endotracheal intubation and preparations made for laparotomy. Emergency operation should be carried out if the endoscopist cannot stop the bleeding and there is no option for IR with angioembolisation or infusion therapy [12].

Up to 15 % of endoscopically treated patients experience recurrent bleeding [13]. Pre-endoscopic predictors of rebleeding are haemodynamic instability and co-morbid illness. Endoscopic predictors of rebleeding are active bleeding at endoscopy, large ulcer size, posterior duodenal ulcer and lesser gastric curvature ulcer [13].

Several studies have looked at patients at high risk of rebleeding and whether they should have early elective or planned operation. A randomised trial of patients with initial endoscopic control of ulcers with arterial bleeding or large visible vessel found that 4 % of the surgery group rebleed versus 50 % in the repeat endoscopic treatment group although the mortality rate was the same [14]. Other prospective studies have shown that early planned surgery in high-risk groups may be beneficial [15]. It is likely that the endoscopic techniques in these studies are not current best practice and there would appear to be no place for routine operation or IR in patients who have had successful endoscopic haemostasis. Similarly most authorities do not recommend routine second-look endoscopy although it should be considered in high-risk patients [4].

For patients that do rebleed after initial successful endoscopic therapy, there is evidence from a randomised trial that repeat endoscopy is superior to surgery. In that study, endoscopy reduced the need for operation without increasing the risk of death, and there were fewer complications [16]. It is generally accepted that a second attempt at endoscopic haemostasis is indicated although there may be a subgroup where surgery may be a better option such as large posterior duodenal ulcer and shock at first presentation [15]. If available, IR has a role in this group and appears superior to surgery. A retrospective study of 118 patients for endoscopy refractory ulcer bleeding found it to have a lower mortality and lower rate of complications. Surgery was associated with a higher rate of primary haemostasis (100 % vs 91 %, p = .007) and a lower rate of rebleeding (15 % vs 40 %, p = .004) but also a higher rate of complications (60 % vs 38 %, p = .02) and an increased mortality when adjusting for confounding factors [17].

7.4 Surgery

The role of surgery in UGIH is now restricted to patients who fail primary endoscopic haemostasis or rebleed after having had two therapeutic endoscopies and where IR is not available or has itself failed. Endoscopic failure of haemostasis occurs in 5–10 % of patients due to reasons such as inability to get a view due to blood and clots, the ulcer position or that the vessel is too large to control with the available equipment and expertise [17]. In the rare case of a patient presenting with exsanguinating haemorrhage, urgent operation will be needed but usually an on table endoscopy is attempted to at least exclude varices.

Fig. 7.3
figure 3

Aortoenteric fistula in the distal duodenum due to knitted polyester (Dacron) graft which had ulcerated through the intestinal wall. The patient presented with melaena and the first endoscopy had not detected the problem as the surgeon only went to the second part of the duodenum. Once a diagnosis of aortoenteric fistula is entertained, discussion with a vascular surgeon is mandatory, and emergency surgical intervention will usually be required. In most instances, the aortic graft is removed with closure of the duodenum, followed by an extra-anatomic vascular bypass to revascularise the lower extremities. The perioperative mortality rate is high, and major complications are common. Lesser alternative procedures such as not removing the graft primarily or endovascular stenting as a bridge to more definitive treatment in a high-risk patient have been described [26]

Full size image

While surgery plays a minor role in the overall management of the community of patients with UGIH, the surgery itself can be a significant challenge. Most emergency surgeons nowadays will not have had experience with elective surgery for peptic ulceration. This is then coupled with the fact that the cases that come to emergency surgery will tend to be difficult ulcers in poor-risk patients due to chronic co-morbidities who are acutely unwell after numerous endoscopic attempts at haemostasis over several days. Surgeons should intervene decisively and the operation should be tailored but, in principle, kept to the simplest procedure possible, i.e. conventional open surgery and underrun of a bleeding vessel.

Operations for UGIH today focus on safely arresting the haemorrhage and there is no place for acid-reducing operations sometimes called definitive surgery. Historically the surgical options for a bleeding duodenal ulcer included vagotomy (truncal or selective) and drainage procedures or gastric resection. Each of those operations was associated with an incidence of ulcer recurrence, postgastrectomy syndrome and mortality. With the availability of effective acid-suppressing drugs and knowledge of the role of H. pylori, logically there is no need for the surgeon to do a definitive acid-reducing procedure. Occasionally when operating for a chronic bleeding ulcer, a distal gastrectomy may be necessary, but even then the resection should be limited and undertaken for technical reasons and not for long-term acid control.

There are still surgeons that challenge this change in practice citing a lack of high-quality evidence and still feel there is a place for so-called definitive operations as a routine. In fact, there are published data to support their view, and Schroder and colleagues recently reported a retrospective study of emergency surgery for bleeding peptic ulcers. They found that vagotomy and drainage was superior to local oversew as it was associated with a significantly lower 30-day postoperative mortality rate (of note distal gastrectomy had a higher rate). They postulated that the decreased mortality observed among patients who underwent vagotomy/drainage was due to superior perioperative acid reduction (through vagotomy) and were therefore less likely than patients who underwent local oversew to have recurrent ulcer bleeding in the postoperative period. However, the difference in major postoperative bleeding rates between the two groups did not reach statistical significance and they could not confirm that the patients included in the analysis received appropriate medical therapy (proton pump inhibitors and/or H. pylori eradication) after their emergency ulcer oversew operation [18].

A recent much larger population-based cohort study from Sweden of 4163 patients compared definitive surgery with a minimal approach for refractory peptic ulcer bleeding in relation to survival. This study found overall survival after minimal surgery was no worse and was associated with better long-term survival during the more recent study period and concluded that a minimal approach is sufficient in most cases [19].

While laparoscopic surgery for perforated peptic ulcer is well established, it appears not to have a role at present for UGIH and conventional open surgery is the standard of care when operation is required. A midline epigastric incision is made and if the source of bleeding had not been identified on gastroscopy, inspection and palpation may give a clue as to where the lesion lies. A longitudinal gastrostomy commencing from the pre-pyloric region through the pylorus to the first part of the duodenum is the standard and the incision can be easily extended proximally. If a gastric lesion is likely, then this initial incision does not need to cross the pylorus. Babcock-type forceps or stay sutures can be used to hold open the gastrotomy.

Clots and blood should be removed from the stomach and sometimes narrow Deaver-type retractors can be used to retract the edges of the gastrotomy to get a view of the mucosa especially proximally. A small lesion may be easier to feel than see, and this is classically the case for the Dieulafoy lesion palpable in the gastric fundus. When the lesion is found to be high in the stomach, for example, a Dieulafoy lesion or Mallory-Weiss tear at the OG junction, access can be difficult. A Goligher frame with a sternal hook to elevate the sternum can be used with a Deaver retractor to elevate the left lobe of the liver. It may be necessary to dissect around and mobilise the distal oesophagus, sometimes dividing some upper short gastric vessels taking care with the spleen. This is most easily done with a harmonic scalpel. A nylon tape can be slung around the distal oesophagus and with this the surgeon has control to assess and oversew any proximal lesion.

For bleeding gastric ulcers (GU), a gastrotomy with underrunning of the bleeder with biopsy may be appropriate, while those in a favourable position, such as those away from the lesser curve, local excision, may be a better option. A nasogastric tube is usually left in for 24 h.

The preferred operation today for a bleeding duodenal ulcer (DU) is to underrun the bleeder having exposed the lesion via a longitudinal duodenotomy which usually will have crossed the pylorus. This opening can be closed anatomically or with a pyloroplasty. I have always found that anatomical closure is easier and appears more secure and the theoretical risk of stenosis has not been borne out.

The gastroduodenal artery is usually the vessel visible in a large posterior DU and is ligated proximally and distally to the bleeding site. Numerous authors advocate a third suture as a horizontal mattress placed to control haemorrhage from the transverse pancreatic branch of the gastroduodenal artery. Some surgeons recommend using a non-absorbable suture such as polypropylene. There are some large penetrating posterior DUs or rarely the bleeding and perforated posterior DU that require a distal gastric resection due to the first part of the duodenum having been destroyed. This is a conservative resection done inside the gastroepiploic arcade to enable safe control of the acute pathological problem and not as a definitive antiulcer operation. Gastrointestinal continuity can then be re-established by either gastroduodenostomy (Billroth I) or gastrojejunostomy with closure of the duodenal stump (Billroth II). The former is preferred due to the better functional outcome; however, it can be a more technically challenging operation. Traditionally a drain has been left to the site of the duodenal closure in the Billroth II procedure and left for up to a week. Numerous methods have been described in the past for dealing with the difficult duodenal stump closure including forming a controlled fistula via a tube duodenostomy. These operations have the added advantage of excluding the bleeding vessel/ulcer base from the GI tract so rebleeding cannot occur. A feeding jejunostomy should be made in any high-risk case to enable early nutrition while protecting the suture line.

Millat and colleagues reported a randomised controlled trial comparing vagotomy and pyloroplasty with gastric resection for bleeding DU. They found that the rebleeding rate was higher (17 % vs 3 %) with vagotomy and pyloroplasty, but the overall mortality was not different [20]. Similarly a multicentre randomised prospective trial by Poxon and colleagues compared minimal surgery (underrunning the vessel or ulcer excision and adjuvant ranitidine) with conventional ulcer surgery (vagotomy and pyloroplasty or partial gastrectomy) for the treatment of bleeding peptic ulcer. Sixty-two received conservative surgery and 67 conventional operation. Twenty-nine patients died, 16 (26 %) after conservative surgery and 13 (19 %) after conventional operations. The only significant difference between the groups was the incidence of fatal rebleeding, which occurred in six patients (10 %) after conservative surgery compared with none after conventional surgery (P < 0.02) [21]. These are historical results before the use of PPI and when operations were done to control acid, but they show that gastrectomy, if it can be done safely, is a very effective way to definitively stop bleeding from a difficult DU. Emergency partial gastrectomy in the elderly carries a higher morbidity and mortality rate, but a recent report showed an anastomotic leak requiring reoperation occurring in only 6 % with no recurrence of DU disease [22, 23].

Rebleeding after operation can be a significant management problem. Historically, the rates have been relatively high when simple underrunning of the ulcer, leaving it in the GI tract, was coupled with inadequate acid suppression. Usually endoscopic treatment is not appropriate due to the immature suture line, and IR should be used if at all possible. When reoperation is needed, a more aggressive procedure would usually be taken with distal gastrectomy as described above, and a feeding jejunostomy should be formed.

Patients may require monitoring in an intensive care or high dependency unit depending on their co-morbidities and condition at the completion of the operation. Acid-reducing therapy with PPI should continue intravenously. Consideration should be given to early feeding via a nasoenteric tube or jejunostomy that was placed at the time of surgery. For duodenal ulcers, empiric H. pylori eradication therapy can be used, and for gastric ulcers, a repeat endoscopy with biopsy should be done to exclude an underlying malignancy. Patients who require continuation of dual antiplatelet therapy (e.g. aspirin and clopidogrel) should remain on PPI although management should be individualised in consultation with their cardiologist [4]. Unusual or recurrent acid-peptic lesions require a serum gastrin level (off PPI) to exclude the Zollinger-Ellison syndrome (gastrinoma secreting abnormally high levels of gastrin). Patients who have had gastrectomy will require monitoring of their nutritional state and may require iron and vitamin B12 supplementation.

Uncommon causes of UGIH, i.e. those not due to peptic ulceration, that come to operation have to be dealt with on a case-by-case basis, but the general principles include adequate exposure and assistance, secure haemostasis and distal enteral nutrition.

7.5 Variceal Haemorrhage

Patients with acute variceal haemorrhage from portal hypertension require a different management strategy to those with non-variceal UGIH due to the underlying mechanism of the bleeding. Discussion with or transfer to a liver unit should be initiated. First-line treatment includes resuscitation, correction of coagulopathy or thrombocytopaenia, vasoconstrictor drugs such as terlipressin and endoscopic band ligation. Failure to control bleeding or early rebleeding means second-line treatments should be instituted. These include balloon tamponade and insertion of a transjugular intrahepatic portosystemic shunt (TIPS), and recently there are promising reports of insertion of removable covered self-expandable metallic oesophageal stents. Balloon tamponade achieves haemostasis by direct compression of the bleeding varices and is effective in 80 % of cases but requires close monitoring due to its high complication rate [24, 25]. In-depth discussion of the management of this complex problem is beyond the scope of this chapter.