Approach to the management of acute esophageal variceal bleeding (EBL, endoscopic band ligation; TIPS, transjugular intrahepatic portosystemic shunt)
Initial Assessment and Resuscitation
In patients presenting with upper gastrointestinal bleeding, history and physical examination findings suggestive of chronic liver disease, together with laboratory parameters (e.g., thrombocytopenia) and/or imaging features (e.g., splenomegaly, shrunken nodular liver, presences of intra-abdominal varices) suggestive of portal hypertension and cirrhosis, are clues to a potential case of variceal bleeding. In patients with decompensated cirrhosis, other cirrhotic complications, such as hepatic encephalopathy (HE), spontaneous bacterial peritonitis (SBP), or hepatorenal syndrome (HRS), may also accompany an episode of acute EV bleeding [1, 7, 16–18].
The airway should be assessed and secured in patients presenting with hematemesis. This is particularly important in encephalopathic patients who are at high risk for aspiration of blood and gastric contents. Intensive care unit admission should be considered for EV bleeding patients with hemodynamic instability and significant HE requiring intubation before proceeding to endoscopy.
Resuscitation with blood volume and fluid replacement should be implemented promptly with the goals of maintaining hemodynamic stability and a hemoglobin level around 8 g/dL [16, 17]. Caution should be taken not to over-transfuse or over-volume expand as variceal rebleeding may be precipitated based on data from animal studies showing that restitution of all lost blood would lead to an increase in portal pressure, resulting in more rebleeding and mortality [16, 17, 20, 21]. In a recent large randomized trial that compared a restrictive transfusion strategy (transfusion to 7 g/dL) to a liberal transfusion strategy (transfusion to 9 g/dL) in 921 patients with acute upper GI bleeding, a lesser rebleeding rate and higher survival at 6 weeks were observed in the restrictive transfusion group . The difference in survival was mainly noted in patients with Child-Pugh class A and B cirrhosis, providing evidence that a restrictive transfusion approach may be beneficial in patients with variceal bleeding .
In patients with significant coagulopathy and thrombocytopenia, transfusion of fresh frozen plasma and platelets, respectively, can be considered [16, 17]. Two randomized controlled studies evaluated the utility of recombinant factor VIIa [23, 24]. Both studies did not find a beneficial effect of recombinant factor VIIa over standard therapy, although a post hoc subgroup analysis showed a reduced proportion of Child-Pugh class B and C patients with failed control of variceal bleeding [23, 24]. Currently, there is insufficient evidence to support the routine use of recombinant factor VIIa in the management of acute variceal bleeding.
Cirrhotic patients who develop upper gastrointestinal bleeding are at high risk for bacterial infections, such as SBP, which in turn put them at greater risk for variceal rebleeding and higher mortality [10, 11]. The use of short-term antibiotic prophylaxis up to 7 days in cirrhotic patients with variceal bleeding, with or without ascites, has been shown to reduce the risks of bacterial infections, variceal rebleeding, and mortality [25–28]. Quinolones can be used in most patients, while ceftriaxone has been reported to be more effective in patients with Child’s class B or C cirrhosis or in centers with high quinolone-resistant organisms [16, 17, 29].
When variceal bleeding is suspected, administration of vasoactive drugs should be initiated as soon as possible before endoscopy [16, 17]. Vasopressin, terlipressin, and somatostatin and its analogs (e.g., octreotide or vapreotide) have been investigated in the treatment of acute variceal bleeding [16, 17]. Meta-analyses of more than 15 trials have shown that vasoactive drugs are comparable to emergent endoscopic sclerotherapy as initial therapy for variceal bleeding with regard to outcome measures of rebleeding and mortality but with fewer adverse events [30, 31].
Vasopressin is a potent splanchnic vasoconstrictor and is effective in the control of acute variceal bleeding . However, its clinical utility is often limited by its unfavorable side effect profile, including bowel, cardiac, and peripheral ischemia . The addition of nitrates helps to reduce the side effects of vasopressin monotherapy, but overall side effects are still higher than those associated with terlipressin and somatostatin analogs [16, 17, 32–34]. Hence, the use of vasopressin is generally limited to a maximum of 24 h in order to minimize its side effects .
Terlipressin is a synthetic analog of vasopressin with a longer biological activity and significantly fewer side effects. In one randomized study, early administration of terlipressin with glyceryl trinitrate improved bleeding control and mortality . Data from a meta-analysis of 7 studies have demonstrated that terlipressin reduces failure of hemostasis and mortality when compared to placebo . When compared to emergent endoscopic sclerotherapy, terlipressin has been shown to have similar efficacy in terms of bleeding control and mortality . Terlipressin can be initiated at a dose of 2 mg every 4 h. Once bleeding is controlled, it can be titrated down to 1 mg every 4 h for up to 5 days to prevent rebleeding [16, 37]. The efficacy of terlipressin for control of acute variceal bleeding ranges from 75 to 80 % at 48 h and 67 % at 5 days [36, 37]. Severe side effects, such as peripheral and cardiac ischemia, were reported in less than 3 % of treated patients .
Somatostatin causes splanchnic vasoconstriction by inhibiting the release of vasodilatory hormones . Somatostatin is given as an initial bolus of 250 μg, followed by intravenous infusion at 250 μg per hour for up to 5 days . Somatostatin has been shown to be as effective as emergent endoscopic sclerotherapy, but with fewer complications . In patients undergoing endoscopic sclerotherapy, early administration of somatostatin was demonstrated to be more effective than placebo in the overall control of acute variceal bleeding .
Octreotide is a somatostatin analog with a longer half-life that causes splanchnic vasoconstriction by inhibiting the release of vasodilatory peptides and by local vasoconstrictive property . It is usually administered as an initial bolus of 50 μg, followed by an infusion of 50 μg per hour for up to 5 days . While octreotide was shown to be equally effective in the control of initial bleeding and rebleeding rate when compared to emergent endoscopic sclerotherapy in an early study, there has been some controversy about its efficacy as single therapy [16, 40]. It has been postulated that rapid development of tachyphylaxis may be the reason behind the inconsistent results found in studies using octreotide alone [16, 41]. The benefit of using octreotide as an adjunctive therapy in patients who have undergone endoscopic therapy for EV is more evident . Results from a meta-analysis have shown that octreotide reduces rebleeding in patients treated with endoscopic therapy . Both terlipressin and octreotide are similarly efficacious as adjunctive therapy to endoscopic therapy in patients with variceal bleeding . Octreotide may be the vasoactive drug of choice in countries where terlipressin is not available.
Timely endoscopy (within 12 h of presentation) should be performed in patients with suspected EV bleeding [16, 17]. Bleeding from EV is confirmed when active (Fig. 12.2) or definite stigmata of recent hemorrhage, such as fibrin clot/nipple sign (Fig. 12.3), are identified at endoscopy. If definite variceal bleeding stigmata are not seen at endoscopy, upper GI bleeding can be attributed to EV in the absence of other sources of hemorrhage, particularly when the varices are large and demonstrate red signs, such as red wale markings (Fig. 12.4).
Esophageal varix with active bleeding (arrow)
Esophageal varix with fibrin clot/nipple sign (arrow)
Esophageal varices with red wale markings (arrows)
Options for endoscopic therapy include endoscopic band ligation (EBL), endoscopic injection sclerotherapy, and endoscopic variceal obturation with a tissue adhesive (e.g., cyanoacrylate).
Endoscopic Band Ligation
Several multiband ligation devices are available that can fit both standard and therapeutic channel gastroscopes. If available, a therapeutic channel gastroscope is recommended due to improved suction capability even with the banding device in place. Placement of up to 6 bands per treatment session is generally sufficient. Placement of >6 bands per treatment session does not improve bleeding or variceal eradication outcomes and is associated with increased overall procedure times and misfired bands.
In a patient who did not undergo endotracheal intubation for airway protection, the procedure is performed with the patient in the left lateral decubitus position and the head of the bed raised about 30° to minimize the risk of aspiration. In the endotracheally intubated patient, the procedure can be performed either in the supine or left lateral decubitus position.
Once in the esophagus, the EBL-loaded endoscope is oriented toward the varix and suction is applied. A large varix can be readily suctioned into the cap, whereas a varix that is small or associated with scarring from prior band ligation may require gentle back and forth movement of the shaft of the endoscope or right-left ratchet manipulation with continuous suction to entrap enough of the varix into the cap for effective ligation. Ideally, a “red out” is obtained during suction of the varix into the cap prior to band deployment. If the band is deployed prior to suctioning enough of the varix into the cap, it will most likely slip off and result in a misfire, as well as inciting trauma-induced bleeding. Unless the esophageal lumen is capacious, passage of the endoscope distal to the banded varix should be avoided as this may result in friction and band dislodgement.
In the presence of an actively bleeding varix, the first band should be placed immediately onto the bleeding site (Video 12.1). If the initial ligated varix is located at or near the gastroesophageal junction (GEJ), additional bands can be placed in a cephalad and spiral fashion, targeting the varices in the distal 5–7 cm of the esophagus. If the index bleeding site is in a more proximal location (e.g., proximal or mid-esophagus), it may be preferable to forego placement of further bands distal to the index band since the latter may become dislodged from friction generated by passage of the endoscope, resulting in rebleeding (Video 12.2). If bleeding is torrential, the bleeding varix may be difficult to pinpoint. In this situation, EBL of varices at the cardia or GEJ may be considered, as this often allows decompression of flow so that bleeding from the offending varix can be slowed, identified, and ligated.
The presence of a nipple sign or fibrin plug confirms the site of variceal hemorrhage and should be targeted for EBL (Fig. 12.5). If the target varix with stigmata of recent hemorrhage is more proximally situated, band ligation of varices can start at the GEJ and conclude with the culprit varix.
(a) Endoscopic band ligation targeting the bleeding point of an esophageal varix indicated by the presence of fibrin clot/nipple sign. (b) Banded varix
With regard to follow-up, EBL is typically repeated at 2–4-week intervals until variceal eradication is achieved. EBL at shorter time intervals may be problematic since post-banding ulcers may still be present and interfere with placement of bands.
Endoscopic Injection of Sclerosants and Cyanoacrylates
Several sclerosing agents have been used for variceal sclerotherapy with comparable efficacy (Table 12.1). The injection sites are sealed off due to the local pressure effects of the sclerosants and thrombosis of the varices secondary to inflammation. The selection of a particular sclerosant is largely dependent on operator preference and availability. A freehand injection technique using a 23-gauge injection needle is typically employed for variceal sclerotherapy. Injection of the sclerosant can be performed directly into the varix (intravariceal) or adjacent to the varix (paravariceal), with resultant edema and blanching of the injected area (Fig. 12.6). When feasible, intravariceal injection is preferred since it is associated with fewer complications than the paravariceal approach, although the intended injection technique may not be achieved when the field of view is obscured by blood. In the setting of active variceal bleeding or stigmata of recent hemorrhage, injections are directed in and around the bleeding site (Video 12.3). The injection volume depends on the type of sclerosant utilized and size of the varix (Table 12.1).
Volume per injection site (ml)
Maximum volume per session (ml)
Relative tissue injury
Fatty acid derivatives
Ethanolamine oleate, 5 %
Sodium morrhuate, 5 %
Sodium tetradecyl sulfate, 1 % and 3 %
Polidocanol, 0.5–3 %
Ethanol, 99.5 %
Phenol, 3 %
(a) Salvage sclerosant injection into a bleeding varix for which banding failed due to surrounding scarring from prior band ligation. (b) Swelling and blanching of the bleeding varix during sclerotherapy, resulting in hemostasis
Variceal obturation refers to the injection of a tissue adhesive (e.g., cyanoacrylate) directly into the variceal lumen, causing obliteration of the varix. Injection of cyanoacrylate is less commonly performed for bleeding EV compared to bleeding GV, although the injection technique is relatively similar. The technical aspects of cyanoacrylate injection for variceal hemorrhage are described in a separate chapter.
Data from multiple randomized trials support the use of EBL as the preferred endoscopic treatment for acute EV bleeding, with less rebleeding and less complications in the EBL group when compared to the sclerotherapy group [45–52]. Two studies reported a lower mortality rate in the EBL group [45, 52]. In a study of patients with actively bleeding EV at endoscopy, EBL and sclerotherapy were comparable in achieving hemostasis of oozing varices, but EBL was superior to sclerotherapy in the control of spurting varices . Injection sclerotherapy is an alternative in patients in whom EBL is not feasible (e.g., in patients with extensive scarring from prior EBL). Endoscopic ultrasound (EUS) may have a role in monitoring EV eradication and provide prognostic information after index EBL . In patients who underwent EBL for EV bleeding, the presence of large paraesophageal varices at EUS within 4 weeks of index EBL was shown to predict recurrence of EV and rebleeding .
While extensive literature is available regarding the use of cyanoacrylate injection for obturation of bleeding gastric varices (GV), data are scant regarding the use of cyanoacrylate injection for bleeding EV [16, 17]. Limited data from one small prospective case series and two small randomized studies reveal conflicting results between EBL and cyanoacrylate injection in terms of acute control of EV bleeding, rebleeding rate, and mortality in patients with cirrhosis [54–56]. Of note, cyanoacrylate injection for bleeding EV is not approved for use in the United States, although it has been utilized on an off-label compassionate basis when standard therapies have failed.
The combination of vasoactive drugs and endoscopic therapy is the current standard of care for acute EV bleeding [16, 17]. Results from a meta-analysis of 8 trials showed that combined pharmacological and endoscopic therapy improved control of initial bleeding and 5-day hemostasis when compared to endoscopic therapy alone (EBL or sclerotherapy) . No difference in severe adverse events or mortality was found between the two groups .
Despite the use of pharmacological and/or endoscopic therapy, variceal bleeding may not be controlled in about 10–20 % of cases [16, 17]. Rescue therapies, such as balloon tamponade, transjugular intrahepatic portosystemic shunt (TIPS), shunt surgery, and self-expandable metal stents, are potential options in selected patients, depending on the liver disease status and comorbidities.
Balloon tamponade is considered a temporary bridge to more definitive therapy for control of variceal bleeding when primary endoscopic hemostasis fails [16, 17]. Several designs of equipment are available: Sengstaken-Blakemore tube, Minnesota tube, and Linton-Nachlas tube. The patient should be admitted to the intensive care unit with intubation for airway protection before insertion of the balloon tamponade system. Initial control of variceal bleeding was reported to be successful in more than 80 % of patients . Because severe complications including aspiration pneumonia, esophageal perforation, tissue necrosis, and acute laryngeal obstruction are not uncommon, balloon inflation generally should not exceed 24 h . Rebleeding rate is high after balloon deflation, and thus, a more definitive therapy (e.g., TIPS) should be planned within 24 h of balloon inflation.
Transjugular Intrahepatic Portosystemic Shunt
TIPS has been accepted as one of the salvage therapies for unsuccessful endoscopic control of variceal bleeding when local expertise is available. The efficacy of TIPS for uncontrolled EV bleeding despite pharmacological and emergent endoscopic therapy has been demonstrated in multiple studies [59–63]. Control of hemorrhage is achieved in more than 90 % of patients [59–63]. In a small study of patients with HVPG >20 mmHg, the early use of TIPS was reported to improve survival . However, patients with decompensated cirrhosis may not benefit from TIPS. In patients with poor liver reserve and multi-organ failure, the 30-day mortality could reach 100 % [60, 61]. In candidates who are fit for TIPS placement, hepatic encephalopathy and TIPS stenosis over time have been two major concerns following TIPS. Hepatic encephalopathy occurs in 25–35 % of patients after TIPS placement . Patients who develop severe encephalopathy may require either a narrower stent to reduce the size of the shunt or total occlusion of the initial stent. Stent occlusion or stenosis by thrombosis occurs in about half of the patients within 2 years after TIPS creation when bare metal stents are used . Newer polytetrafluoroethylene (PTFE)-covered stents have been reported to increase TIPS patency, with patency rate greater than 70 % at 2 years [67, 68]. In a randomized control trial comparing early TIPS (within 72 h after presentation of variceal bleeding) to continuation of vasoactive therapy and insertion of TIPS as a rescue therapy, the early use of TIPS was associated with a significant reduction in treatment failure and improvement in survival . Therefore, early TIPS should be considered in these patients.