Bleeding from oesophageal varices is the most common cause of serious gastrointestinal haemorrhage in children. Bleeding may occur at any age, but some patients with varices never bleed.1 The risk of bleeding is not linearly related to portal pressure, but to the size of the varix and the thickness and integrity of its wall.2 Thus varices are most likely to bleed if they project prominently into the oesophageal lumen, if the overlying mucosa is blue, and particularly if there are ‘cherry red spots’ on the varix. Salicylate ingestion used to be recognised as an important precipitant.

The treatment of bleeding oesophageal varices is dependent on the underlying cause. In patients with portal hypertension fromintrahepatic liver disease treatment is dictated by the latter and may determine the need for liver transplantation. Patients with good liver function and bleeding varices can, however, be successfully managed by treatment of their portal hypertension alone. Opinions on the primary management of extrahepatic portal hypertension have long been divided between those who advocate portosystemic shunting and those who favour endoscopic injection sclerotherapy. The results from studies of large series of children undergoing endoscopic injection sclerotherapy have encouraged the widespread acceptance of this technique in children with intrahepatic disease, in whom prognosis is determined more by underlying liver pathology, and in those with portal vein thrombosis or presinusoidal venous obstruction in whom variceal bleeding is the main threat to life.3-5

The management of acute variceal bleeding involves prompt but careful resuscitation. Shock should be corrected by cautious blood transfusion, but over transfusion may increase splanchnic blood flow and precipitate rebleeding. Coagulopathy and severe thrombocytopenia (platelet count less than 50 × 10⁹/l) require correction and sepsis should be sought and treated with intravenous antibiotics. Patients must be monitored closely for continuing haemorrhage and the development of hepatic encephalopathy, and arrangements should be made for transfer to a centre where flexible fibreoptic endoscopy can be carried out by an operator skilled in injection sclerotherapy.

The use of somatostatin and its longer acting synthetic analogue octreotide has not been fully evaluated in children. In some adult studies these drugs have proved to be as effective as emergency sclerotherapy6 7 and balloon tamponade8 in controlling bleeding. Moreover, they are easily administered by continuous intravenous infusion and serious side effects are very rare. They may be particularly useful in stabilising a child before transfer or in preventing early rebleeding after sclerotherapy.9 As they are not effective in non-variceal gastrointestinal bleeding,10 diagnostic upper gastrointestinal endoscopy is essential.

Flexible fibreoptic endoscopy is ideally carried out under general anaesthesia with an endotracheal tube in situ within 24 hours of presentation. After assessment of the varices and the upper gastrointestinal tract, the varices are injected using a flexible endoscopic needle. Injections are concentrated at the cardia and in the lower 3 cm of the oesophagus (the site of perforating veins) and are predominantly intravariceal. The details of the technique and type of sclerosant have been reviewed elsewhere.11 Patients are given ranitidine and sucralfate by mouth for up to two weeks after each injection session and antibiotic prophylaxis is recommended for those with damaged/prosthetic heart valves. The initial two or three treatments are performed at weekly intervals, but injection is deferred for one week if significant oesophageal mucosal ulceration is present. Further treatments are given at one to three monthly intervals until the varices have either disappeared or been converted into thrombosed cords. Most varices can be obliterated within one year by five to eight injection sessions.3 Regular checks, usually as a day case, are then carried out every six to 18 months until a stable state is achieved.

The efficacy of sclerotherapy in treating oesophageal varices is clearly shown by the King’s College Hospital series of 108 children treated in the 1980s.3 Endoscopically confirmed obliteration of bleeding oesophageal varices was achieved in all children with portal vein obstruction and 84% of those with intrahepatic disease (for example, biliary atresia, congenital hepatic fibrosis, cystic fibrosis). All but one of the children whose varices were not obliterated did not complete a course of sclerotherapy because of liver transplantation or death from liver failure. Only one child died from variceal haemorrhage.

Injection sclerotherapy has been associated with numerous complications,11 but these are rarely serious in children. Transient fever and mild retrosternal discomfort are common sequelae after injection and are usually self limiting, but may signify bacteraemia. Gastrointestinal bleeding before variceal obliteration is complete occurs in 40% of patients3 and is usually due to a non-thrombosed varix or an oesophageal mucosal ulcer, but may be secondary to peptic ulceration. The short term incidence of recurrent oesophageal varices and bleeding from gastric varices is low (<10%) and the former often respond to further sclerotherapy.3 In adults, sclerotherapy ulcers can be reduced by prophylactic ranitidine12 and the incidence of oesophageal strictures can be decreased by using smaller volumes of sclerosant and prophylactic sucralfate.13 Some children develop oesophageal dysmotility and gastro-oesophageal reflux, which may cause intermittent dysphagia and heartburn. Systemic dissemination of the injected sclerosant leading to distant complications has been reported, but appears to be rare.11 14

Little is known about the long term outcome of children after injection sclerotherapy for bleeding oesophageal varices. In those with cirrhosis, the long term outcome is jeopardised not only by the complications of portal hypertension, but also by deteriorating liver function. Treatment of bleeding oesophageal varices by sclerotherapy is effective in controlling haemorrhage and, unlike portosystemic shunting, does not reduce portal perfusion or carry the risk of encephalopathy. Moreover, it does not add to the technical difficulties of liver transplantation provided sclerotherapy is not carried out immediately before surgery. Those with serious underlying liver disease are, however, often best managed by liver transplantation. For children with extrahepatic portal hypertension, indications for surgery include uncontrollable variceal bleeding, bleeding from gastric or ectopic varices not accessible to sclerotherapy, lack of access to expert sclerotherapy, and severe symptomatic hypersplenism.15 In one study of 32 children with portal vein thrombosis followed up for a mean period of nine years after endoscopic variceal obliteration, recurrent variceal bleeding developed in 10 (31%) patients, but half of these were effectively controlled by further sclerotherapy.14 Gastric variceal bleeding not amenable to sclerotherapy necessitated portosystemic shunt surgery in four (13%) patients. Two patients required a splenectomy for symptomatic hypersplenism. Endoscopic injection sclerotherapy alone was safe and effective in controlling variceal bleeding from portal hypertension in the long term in over 80% of children with portal vein occlusion.

Two new treatments have affected the management of bleeding oesophageal varices, but their precise role in children has yet to be determined. The first is variceal band ligation, which has been adopted by many as the preferred method of endoscopic control of bleeding oesophageal varices in adults with cirrhosis. It achieves variceal obliteration more quickly, with a lower rebleeding rate than injection sclerotherapy, and complications such as oesophageal stricture may be reduced.16 17 It is no more effective than sclerotherapy in controlling active haemorrhage, however, and recurrent varices may be more common. Preliminary clinical experience in children18 19 suggests that the technique is possible in those older than 2 years, but it is currently more cumbersome than sclerotherapy.

The transjugular intrahepatic portosystemic stent shunt is the second novel approach, whereby a stent is inserted between the portal and hepatic veins within the liver. This can be useful as a temporary measure in selected patients with complicated portal hypertension who are waiting for liver transplantation, but long term patency rates are unknown and reports to date simply suggest that it is a feasible option in some children.20

The role of β blockers in children with portal hypertension remains uncertain. These drugs reduce portal venous inflow by decreasing cardiac output and by blocking β receptors in the splanchnic vascular bed. In adults, trials with propranolol in patients who have already bled from varices have yielded mixed results and if there is a benefit, it is likely to be modest.21 Uncontrolled studies in children do no more than point to a possible role.22Although β blockers may be of value in primary or secondary prophylaxis, there is no good evidence to support prophylactic injection sclerotherapy.11