Management of an UGIB should begin before a diagnosis is made. Although many UGIBs will stop spontaneously, the clinicians treating a patient with an UGIB should assume it will not. All patients with an UGIB should be managed initially by following the fundamentals of resuscitation of shock. The patient’s capacity to maintain a safe airway should be established and if in doubt, the patient should be intubated. The patient’s cardiovascular status and the presence or absence of shock should be rapidly investigated and treated. If the patient has any signs of hemodynamic instability or visible evidence of large volume blood loss, the patient should have large-bore vascular access placed and the blood bank should be alerted for possible unmatched blood needs and a type and crossmatch for packed red blood cells. The physical exam is not as helpful in UGIB as it can be in other disease processes. Signs of shock, stigmata of liver disease, presence of peripheral vascular disease, and evidence of previous surgery can usually be investigated. If gross blood or melena is not obvious, a digital rectal exam and fecal occult blood test should not be overlooked. Bloodwork should include a complete blood count, a coagulation panel, a comprehensive metabolic panel and a blood type and screen, at minimum. If there is any history that suggests the patient will have abnormal coagulation studies, dysfunctional, or absent platelets, fresh frozen plasma and platelets should be made available as well [21].

A complete blood count is useful for ruling out blood dyscrasias, and the presence or absence of thrombocytopenia. The hemoglobin level is useful; however, a trend in hemoglobin levels over time is usually more helpful depending on the clinical situation. The metabolic panel will be most helpful in assessing the patient’s current level of renal or hepatic dysfunction, if any. Coagulation panels are useful in assessing the current level of coagulation impairment and should be correlated to medication history. If a patient is coagulopathic and not taking an anticoagulant, malnutrition, DIC, and underlying liver disease should be urgently considered and investigated. Consideration should be given to ordering platelet function assays in patients taking platelet-altering medications. There is some data to recommend platelet transfusions to normalize function assays in patients taking platelet inhibitors who are having a significant bleeding problem [22]. This has not been studied in the setting of UGIB but deserves consideration in the patient with a massive or hemodynamically significant bleed.

Concurrent to these events, as much history as possible should be elicited from the patient, the family, and any other physicians involved with the patient. In addition to the events surrounding the UGIB, attention should be paid to any previous history of peptic ulcer disease or other forms of GIBs. Also, any history of liver disease, renal disease, malignancy, and cardiovascular disorders will be important as each of these has an implication for the etiology and management of the UGIB. A medication history is vital and must include not only prescription medications, but also over-the-counter medications. Patients frequently have more difficulty identifying the type and quantity of over-the-counter drugs they are taking and often do not know the importance of their NSAID use, for example, in their diagnosis and management. The past surgical history is vital to obtain. The success of any possible intervention in these patients in particular relies on a clear understanding the pre-procedural anatomy. Any prior surgery of the foregut, hepatopancreaticobiliary system, aorta, peripheral vasculature, and other abdominal organs will alter the diagnostic and technical approach of the endoscopist, interventional radiologist, and acute care surgeon.

The diagnostic pathway in an UGIB frequently overlaps with the treatment pathway but the first step is identifying the location of the blood loss as proximal or distal to the ligament of Trietz. Every step taken to localize and characterize the source of the bleed improves the chance of successful treatment. A common tool to differentiate an UGIB from lower gastrointestinal bleeding (LGIB) is placement of a nasogastric tube and evaluation of the aspirate. The conventional wisdom suggests a nasogastric aspirate which returns bilious, but not bloody, fluid should prompt a workup for a LGIB. The logic underlying this step is that an aspirate containing bile reflects the fluid composition in the duodenum as well as the stomach; thus, bilious non-bloody fluid should rule out a source proximal to the ligament of Trietz. Unfortunately, this intuitive step has little evidence to support its use. Palamidessi et al. reviewed the literature for studies correlating findings from a nasogastric aspirate with an esophagogastroduodenoscopy (EGD) in patients with melena or hematochezia without hematemesis. They found only three studies that fit their criteria. Among these the sensitivity and specificity of a positive NGT aspirate ranged from 42 to 84% and 54 to 91%, respectively, with the highest values in a cohort of inpatients being treated for a myocardial infarction when they began to have symptoms [23]. They conclude that a conversation with the consulting gastroenterologist concerning the value of an NGT may be worthwhile. If an endoscopist has assessed the patient and their plan to perform upper or lower endoscopy will not be affected by the results of an NGT aspirate, then foregoing this step may be reasonable. Otherwise, an NGT aspirate may still be considered helpful and warranted.

After the initial resuscitation is underway, the source and etiology of the UGIB must be localized. The best initial procedure is usually endoscopy. EGD has proven to be the most important initial procedure. This modality in skilled hands will also offer several therapeutic options. Localization of the source of the bleed is critical. The success of all endoscopic, angiographic, and surgical interventions depends upon localization of the bleeding source. At the minimum, and with the worst conditions, an EGD can differentiate an UGIB from a LGIB. Should the bleeding fail to stop spontaneously or be controlled endoscopically, narrowing down the source vessels or tissues as much as possible makes the possibility of a successful angiographic or surgical intervention with the least amount of morbidity much more likely.

Therapeutic options for the endoscopist have expanded considerably in the last two decades. The techniques used will depend upon the skill and comfort of the endoscopist and the site and etiology of the bleeding. Options include epinephrine or sclerosant injection, thermal coagulation, hemostatic clip application, banding, or a combination of these techniques. Multiple trials have demonstrated that all of these modalities have similar efficacy in arresting bleeding, preventing rebleeding, and reducing the need for urgent surgical intervention. Success after initial treatment can be as high as 98% [4, 24]. Additionally, biopsies can provide useful information and in some institutions, H. pylori testing from biopsied tissue can be done with results much faster than serology or other types of testing. Neoplastic causes of bleeding can also be established by biopsy and can play an important role in treatment strategies. Furthermore, in the case of transpapillary bleeding, endoscopic retrograde cholangiopancreatography (ERCP) can be helpful for both localization and treatment. Recurrent bleeding after endoscopic control is infrequent, occurring in <10% of patients in centers with strong endoscopy departments. Repeat endoscopy for bleeding after initial endoscopic control has been compared to surgical intervention and has been demonstrated to achieve long-term hemostasis in a majority of patients while avoiding the complications of surgery [24]. Repeat endoscopy is widely regarded as the appropriate response to evidence of rebleeding after initial endoscopic control [21].

Several studies have shown a role for high resolution multidetector CT (MDCT) scan with digital subtraction angiography in localizing GI bleeding, including UGIBs. The “pooled” sensitivity and specificity of MDCT for GIB are 86% and 95%, respectively [25]. In swine and in vitro models, high quality MDCT imaging has been shown to localize bleeding as slow as 0.35–0.5 ml/min [26]. MDCT is currently considered “usually appropriate” by the American College of Radiology as the next diagnostic step following a negative or inconclusive endoscopy [4]. MDCT may be the best next step in a patient in whom an aortoenteric fistula is suspected or in a postsurgical patient in whom the anatomy is unclear. Additionally, if complications of neoplastic disease are high on the differential diagnosis, a CT scan may be highly valuable in assessing the extent of disease and these data could dramatically change further diagnostic and therapeutic plans. Unfortunately, at this time MDCT offers few possible concurrent therapeutic maneuvers and remains a solely diagnostic modality.

Technetium-99 m-labeled erythrocyte scans, commonly known as tagged red cell scans, are useful because they can detect bleeding as slow as 0.05–0.1 ml/min. The major drawback to this option is its limited value in localizing the bleeding site. In the case of a LGIB, the relatively immobility of the colon can offer adequate localization for the next therapeutic endeavor. If the source of bleeding is in freely mobile viscera such as the majority of the small bowel, this information is of limited value for any intervention. Surprisingly, given the relatively predictable placement of the stomach and duodenum, studies have shown that tagged red cell scan localization of bleeding lesions in these areas to be frequently inaccurate [27, 28]. Thus, for the UGIB, tagged red cell imaging is rarely appropriate.

With the exception of the discovery of the role of and treatment for H. pylori in peptic ulcer disease, nothing has revolutionized the treatment of GI bleeding as much as advances in interventional radiology in the last two decades. Initially, angiography was available as a purely diagnostic tool and still plays an important role in that respect. Diagnostic angiography can detect bleeding at rates as low as 0.5 ml/min. Although it is generally not as helpful as endoscopy in characterizing the cause of bleeding, angiography can provide anatomic or structural information about lesions which are bleeding at the time of angiography or those that appear to bleed intermittently. Selective visceral angiography is most helpful if the source is arterial; detection of venous bleeding during the venous phase of the imaging is less reliable. Transcatheter arterial embolization (TAE) of UGIB source vessels involves selective visceral angiography and localization of the bleed. This can be expedited by visualization of previously placed endoscopic clips. Thus, hemostatic clip placement at the time of endoscopy is ultimately useful even when the clip placement is not technically successful [29]. Once the site of hemorrhage has been identified, feeding arteries can be embolized with coils, polyvinyl alcohol particles, gelfoam, iso-butyl-2-cyanoacrylate, ethibloc, or a combination of these products. Most studies found no difference in outcomes based on the type of agent used, although no randomized trials have been done to confirm this observation [4, 30]. If the lesion has been at minimum localized to a region of the stomach or duodenum by endoscopy, but cannot be visualized by angiography, “blind” embolization of the vessels feeding this area can be done successfully.

Multiple studies have demonstrated the benefits of TAE in UGIB when endoscopic attempts are not successful. TAE procedural and early clinical success rates (absence of early rebleeding), for bleeding after failure of endoscopic control range from 75 to 100% and 72 to 78%, respectfully in studies published from 2000 onward [20, 31, 32]. This success is highlighted by the fact that many of these studies were done in patients considered too high risk for surgery. The rebleeding rate after TAE in recent studies ranges from 7 to 30%. The factors associated with the risk of rebleeding after TAE have been extensively studied and in multivariate analysis, the presence of coagulopathy or multiple organ failure are consistently associated with clinical failure or rebleeding [20]. The drawbacks of TAE as a treatment option include the requirement for a skilled interventionalist team with 24-h availability, possible inability to cannulate the access vessel secondary to narrow or tortuous vasculature, and inability to localize the bleeding lesion in areas where blind embolization is not an option. Procedural drawbacks include the need for contrast injection and its potential risk of contrast induced renal insufficiency, groin hematomas and pseudoaneurysms, transient hepatic or pancreatic ischemia after embolization, and late duodenal stenosis secondary to embolic ischemia.

There have been no randomized controlled trials comparing TAE to surgery for UGIBs that have failed endoscopic treatment. Several groups have retrospectively compared the two modalities. There are numerous methodological differences amongst these studies; however, most found that patients who underwent embolization rather than surgery were consistently older, had more comorbidities including coronary artery disease, and were more frequently on anticoagulation therapy at the time of the UGIB. Despite this, most showed no significant difference between surgery and TAE in terms of the incidence of recurrent bleeding (23–43%), nor the need for additional surgery [33–36]. This data strongly supports the argument to proceed with angiography and TAE if endoscopic management fails. In cases of transpapillary bleeding, TAE has had excellent results. In studies of TAE in transpapillary bleeding published since 1999, the technical and clinical success rates of TAE have been 88–100% and 75–100%, respectively. The rebleeding rates in these studies ranged from 0 to 21% with major complication rates of 0–3% [20].

The therapeutic options now available for interventional radiologists are various and are not confined to the angiographic approach. They include transjugular intrahepatic portosystemic shunts (TIPS) as well. TIPS insertion for variceal bleeding unresponsive to endoscopic management has excellent outcomes in cessation of bleeding. If combined with variceal embolization, the rate of rebleeding is reduced further. It is equally efficacious as surgical porto-systemic shunts in cirrhotics. Shunts of either type are associated with an increased risk or exacerbation of hepatic encephalopathy. The newer TIPS stents are associated with a secondary patency rate of 90% or higher; additionally, TIPS can be exchanged percutaneously for occlusion and downsized for medically refractory encephalopathy [4].

This approach is most frequently used for a bleeding duodenal ulcer on the posterior wall of the lumen. An upper midline incision is made. The duodenum is mobilized and palpated for the location of the ulcer. An anterior longitudinal duodenotomy is made at the level of the posterior ulcer. The bleeding source is usually an erosion into the lumen of the gastroduodenal artery. Nonabsorbable sutures are used to ligate the vessel proximal and distal to the bleeding lumen. A third “U-stitch” is placed medial to the bleeding area to control inflow from the tranverse pancreatic artery (see Fig. 19.1). All of these sutures must be placed with care and precision in order to avoid any injury to the adjacent common bile duct. Once hemostasis has been achieved, the duodenotomy is closed transversely, most often with the Heineke-Mikulicz pyloroplasty, to avoid any postoperative obstructive symptoms [37, 38].