FOREIGN bodies in the esophagus can cause pressure necrosis of the esophageal wall that may lead to perforation, mediastinitis, and formation of an aortoesophageal fistula (AEF). 1Patients may present with the typical “Chiari triad” of midthoracic pain, a sentinel hemorrhage of bright red blood, and exsanguination hours to days later. 2 

We report the successful resuscitation of a child who had exsanguination from an undiagnosed AEF caused by unsuspected foreign body ingestion. We believe that our patient’s survival was due to immediate, aggressive volume resuscitation and temporary control of the fistula by inflation of an intraoperatively placed esophageal balloon catheter.

A 7-yr-old, 35-kg girl with a history of epilepsy treated with valproic acid presented to an outlying hospital after she had vomited small amounts of dark brown blood. Her only complaints were moderate midepigastric pain on deep palpation. The patient’s hematocrit was 31. After a second emesis of moderate amounts of dark brown blood, she was transferred to Children’s Hospital of Wisconsin (Milwaukee, WI). With vigorous hydration of 45 ml/kg isotonic fluid in the emergency room, her hematocrit dropped to 18. Blood was ordered, a nasogastric tube was placed while the patient was in the emergency room, and saline lavage yielded only scantly blood-stained fluid. A tentative diagnosis of gastric ulcer was made, and she was transported to the operating room for urgent endoscopy without any further routine workup or radiography. Transfusion of 250 ml cross-matched packed erythrocytes was begun in the preoperative area. Suddenly, the child sat up and began to vomit bright red blood profusely. She was rushed to the operating room, and resuscitation with blood and fluids was intensified. Throughout, the child was conscious and followed commands.

A rapid-sequence induction with intravenous ketamine (60 mg) and succinylcholine (40 mg) was performed. The trachea was intubated with a cuffed 6.0-mm-ID endotracheal tube. Midazolam (2 mg) and rocuronium (50 mg) were administered, and the patient underwent ventilation with 100% oxygen. The patient continued to lose massive amounts of blood from the mouth, and despite ongoing fluid resuscitation through two peripheral intravenous cannulae (22 and 20 gauge) in the left and right arms, she sustained a cardiac arrest within minutes. Central pulses were absent, the pulse oximeter failed, and end-tidal carbon dioxide values decreased to single digits, while the electrocardiogram showed a sinus tachycardia of 180 beats/min. Chest compressions were begun and were continued for about 5 min. During 17 min of resuscitation, 900 mg CaCl2, 1 mg epinephrine, 750 ml packed erythrocytes, and 500 ml albumin, 5%, were administered, and an epinephrine infusion (0.1–1.0 μg · kg−1· min−1) was started. These measures resulted in a return of faint femoral pulses and end-tidal carbon dioxide but also in an increase in oral blood loss, which made endoscopy impossible. An emergency laparotomy was performed with the presumption that the bleeding source was gastroduodenal.

The stomach was massively distended with clotted and fresh blood. No source of gastric bleeding was found, but there was a steady flow of bright red blood from the esophagus. After attempted esophagoscopy, a dark, stained 25-cent coin was found in the opened stomach. The esophageal bleeding remained brisk, and a pumping bleeding source was noticed by the endoscopist in the mid esophagus about 18 cm from the incisors. A 24-French Foley balloon catheter was introduced retrograde by the surgeon, and after 15-ml balloon inflation, the bleeding markedly decreased. This measure allowed the anesthesia team to catch up with the blood loss and allowed the surgical team to establish invasive monitoring. The epinephrine infusion was stopped. Another surgical attempt to locate the source of bleeding by deflation of the Foley catheter resulted in very brisk rebleeding and was abandoned.

The cardiothoracic surgeons were consulted. They made a presumptive diagnosis of an AEF and approached the aorta via  left thoracotomy. Extensive friable granulation tissue was encountered between the esophagus and the distal aortic arch. To better visualize the operating field without compromising distal perfusion during aortic cross-clamping, the patient was placed on partial cardiopulmonary bypass with cannulae in the left atrium and the mid descending thoracic aorta. Throughout the whole bypass procedure, the heart continued to beat, providing perfusion to the cerebral and coronary circulations. The patient was fully heparinized (14,000 IU) and was kept mildly hypothermic (34–35°C). Distal perfusion pressure was monitored by a surgically placed femoral arterial line, and proximal pressures were obtained noninvasively by oscillotonometry. A right radial arterial line would have been ideal during partial bypass, but could not be placed. A left radial arterial line was easily placed but was significantly blunted during aortic cross-clamping. A 4-mm AEF was found 0.5 cm distal to the left subclavian artery. The fistula was isolated between clamps placed on the mid transverse arch, i.e. , between the left carotid and subclavian arteries, and the descending thoracic aorta, respectively. The esophagus was closed primarily using a small segment of aortic wall, which was left attached to the esophagus. The diseased segment of aorta was replaced with a 14-mm Dacron graft, presoaked in antibiotic solution. Aortic clamp time was 35 min, and partial bypass time was 40 min. The patient was weaned from partial bypass without difficulty. After closure of the thorax, a feeding gastrostomy tube was placed, and the abdomen was closed. The total volume of fluids and blood products infused was 3,000 ml packed erythrocytes, 500 ml fresh frozen plasma, 3 four-donor units (= 150 ml) of platelets, 1,500 ml albumin (5%), and 4,000 ml lactated Ringer’s solution, totaling an estimated blood loss replacement of 2.5 blood volumes. The patient underwent extubation on postoperative day 2 and was discharged home after 2 weeks. Follow-up clinical examination and esophagography 3 months later were unremarkable.

An AEF is a rare cause of bleeding in the upper gastrointestinal tract. The few cases reported in children were associated with congenital vascular rings or foreign body ingestion. 3–5All reported children with AEF after foreign body ingestion died except for an 8-month-old child in which an extended laparotomy and esophagotomy were performed until the bleeding site was found and manually compressed. 5In our case, the AEF was diagnosed after gastrotomy, and the bleeding was successfully tamponaded with a Foley catheter. The fistula was then repaired through a left thoracotomy using partial cardiopulmonary bypass.

Foreign body ingestion is common in children and may present with atypical symptoms. 3In many cases, parents report suspicion of foreign body ingestion. However, in this child, the bleeding was attributed to a gastric ulcer, and no specific questions were asked. In addition, chest radiography was omitted in both emergency departments. This case suggests that in the absence of a definite cause, a portable chest radiograph should be obtained routinely and early in a child with upper gastrointestinal bleeding. It does not delay further therapy and might yield unexpected results. Placement of a nasogastric tube with the significant risk of dislodging the foreign body and causing additional bleeding clearly would have been avoided.

In the current case, the sequence of events was determined by the profuse bleeding that started in the preoperative area and demanded vigorous fluid resuscitation. We assume that the bleeding slowed down during the arrest period because of tamponade or closure of the AEF with clot or surrounding tissue. After the source of the bleeding was identified, balloon tamponade of the esophagus with a Foley catheter proved to be an effective method to tamponade the bleeding site and allowed us to reestablish an adequate circulating blood volume. Esophageal compression, usually with a Sengstaken-Blakemore tube, has been described in adults for the same purpose and allowed for surgery to close the fistula. 1,6 

Had radiography been performed in this child, the knowledge of a foreign body in the esophagus, together with hematemesis, could have led to the tentative diagnosis of vascular erosion. However, given the rare occurrence of AEF, the most likely next step would have been a rigid esophagoscopy and an attempt to remove the coin. This probably would have led to rebleeding with possible exsanguination as has been described by Gilchrist et al.  3Placement of a balloon catheter from above may have also tamponaded such bleeding. Esophageal balloon tamponade seems to offer the best chance to stop massive arterial bleeding from an AEF so that surgical repair can be performed.

In summary, we report the successful resuscitation and recovery of a child with a previously undiagnosed aortoesophageal fistula. Despite initial exsanguination, massive volume resuscitation, cardiac massage, and esophageal balloon tamponade allowed temporary cardiovascular stabilization of the child and the performance of a definitive surgical repair.

Amin S, Luketich J, Wald A: Aortoesophageal fistula: Case report and review of the literature. Dig Dis Sci 1998; 43: 1665–71
Hollander JE, Quick G: Aortoesophageal fistula: A comprehensive review of the literature. Am J Med 1991; 91: 279–86
Gilchrist BF, Valerie EP, Nguyen M, Coren C, Klotz D, Ramenofsky ML: Pearls and perils in the management of prolonged, peculiar, penetrating esophageal foreign bodies in children. J Pediatr Surg 1997; 32: 1429–31
Sigalet DL, Laberge JM, DiLorenzo M, Adolph V, Nguyen LT, Youssef S, Guttman FM: Aortoesophageal fistula: Congenital and acquired causes. J Pediatr Surg 1994; 29: 1212–4
McComas BC, van Miles P, Katz BE: Successful salvage of an 8-month-old child with an aortoesophageal fistula. J Pediatr Surg 1991; 26: 1394–5
Heckstall RL, Hollander JE: Aortoesophageal fistula: Recognition and diagnosis in the emergency department. Ann Emerg Med 1998; 32: 502–5