Primary Aortoesophageal Fistula due to Thoracic Aortic Aneurysm

Abstract

Aortoesophageal fistula is a rare emergency that presents a real challenge for cardiothoracic surgeons. There have been few reports of survivors. We present the case of a 70-year-old man with aortoesophageal fistula, mediastinal abscess, and severe septicemia consequent to esophageal erosion and rupture of a chronic degenerative descending thoracic aortic aneurysm. The patient underwent successful surgical treatment by aorto–aortic bypass and bipolar esophageal exclusion in conjunction with a cervical esophagostomy and a feeding gastrostomy. The pleural cavity was copiously irrigated and drained. Three months later, a retrosternal gastric bypass operation was performed successfully. The patient's 6-month follow-up examination revealed no problems.

Aortoesophageal fistula (AEF) is a rare cause of gastrointestinal bleeding. Surgery is crucial for survival. It is of great importance that both the thoracic aortic aneurysm and the esophageal erosion be treated in order to avoid massive hemorrhage and septic sequelae.^1,2

Case Report

In January 2006, a 70-year-old man was transferred from a district hospital with the diagnosis of thoracic aortic aneurysm in imminent danger of rupture. The patient had experienced hematemesis as a sentinel event and then presented with loss of consciousness and midthoracic pain. Chest radiography showed mediastinal enlargement, and computed tomography of the chest revealed a 9-cm-diameter aneurysm of the descending thoracic aorta, which compressed the esophagus, and a large thrombus with free air bubbles, which suggested esophageal rupture and mediastinal abscess (Fig. 1).

Fig. 1. Computed tomography shows A) aneurysm of the descending aorta (arrows) and B) air bubbles (white arrow) inside the aneurysm's thrombus that are suggestive of esophageal erosion.

An urgent left posterolateral thoracotomy was performed, with a heart–lung machine on standby. After the intravenous administration of 10,000 IU of heparin, partial occlusion clamps were placed on the aorta below the takeoff of the left subclavian artery and just above the diaphragm; to bypass the aneurysm, a 28-mm Dacron graft was anastomosed end-to-side to the aorta (Fig. 2). Subsequently, the aneurysm was totally excluded and both aortic stumps were closed with continuous 3-0 GORE-TEX® sutures (W.L. Gore & Associates, Inc.; Flagstaff, Ariz) (Fig. 3). The aneurysmal sac was opened, revealing a malodorous mediastinal abscess. The sac was resected, and the thrombus and all infected surrounding tissues were carefully débrided. With difficulty, we then found an esophageal erosion of 2 cm in length. After identifying and ligating the bleeding intercostal vessels, we copiously irrigated and cleaned the pleural cavity with warm normal saline and povidone-iodine solution. The distal intrathoracic esophagus was then divided, just above the diaphragm, with a stapling device. A 2-cm incision was made in the left hemi-diaphragm, and a 10F feeding tube was installed via a gastrostomy. The tube was anchored to the diaphragm and threaded through the chest wall to the exterior. The diaphragm was closed, and two 36F chest tubes were placed in the pleural cavity. After closure of the chest, the patient was placed in a supine position, and, through a left neck incision, the esophagus was divided and brought out as a cervical esophagostomy, with the distal end stapled (Fig. 4).

Fig. 2. Aorto–aortic bypass by means of a Dacron graft that is anastomosed proximally to the descending thoracic aorta below the takeoff of the left subclavian artery and distally just above the diaphragm.

Fig. 3. Exclusion of the descending thoracic aortic aneurysm.

Fig. 4. Final management of the esophageal rupture. The esophagus is divided on each end with stapling devices (the broken line at the diaphragmatic end of the esophagus denotes impending division). The cervical esophagus is brought out as a cervical esophagostomy. Note the feeding gastrostomy tube that is anchored to the left hemidiaphragm and brought out through the chest wall.

The patient was transferred to the intensive care unit. He remained hemodynamically stable and was weaned from the ventilator on the 3rd day. A regimen of imipenem, metronidazole, teicoplanin, and fluconazole was imposed. The chest tubes were removed on the 4th day, but further recovery was complicated by the development of a left pleural empyema consequent to a multiresistant Acinetobacter infection. The chest cavity was properly drained, and the patient was discharged from the hospital a week later on enteral feeding and on antibiotic coverage with ciprofloxacin and clindamycin. The pleural drain was converted to an open empyema tube, which was removed 2 months later, after chest computed tomography showed no anastomotic pseudoaneurysm and the lungs fully expanded, with no residual pleural space or effusion.

Three months after the initial operation, the patient was readmitted for gastroesophageal reconstruction. A retrosternal gastric bypass was successfully performed (Fig. 5). The patient's recovery was uneventful apart from a small esophagocutaneous fistula in the neck due to an anastomotic leak that lasted for 1 month and was treated with enteral nutrition at home, via a nasogastric feeding tube. Six months later, he was well, mobile, and able to swallow adequately.

Fig. 5 A) Chest computed tomogram—combining an aortogram with a contrast-medium swallow image—shows the final reconstruction of the aorta and the upper gastrointestinal tract. The asterisk denotes the proximal end-to-side anastomosis of the Dacron graft (G) to the proximal descending aorta (D). The arrow indicates the retrosternal gastric interposition with a Levin gastric tube inside it and a small amount of contrast medium given orally. B) The line drawing corresponds with the computed tomographic image.

A = ascending aorta

Discussion

Among patients with thoracic aortic aneurysm, rupture is the most common cause of death³; rarely, this erosion penetrates the esophagus as well. Aneurysm of the thoracic aorta is the cause of AEF in more than 50% of the cases, followed by foreign-body perforation of both lumina and by esophageal carcinoma.⁴ By controlling both bleeding and mediastinal contamination, surgery offers the only chance for survival, but there have been few reports of successful management and subsequent outcomes.²

The primary concern must be the repair of the aortic aneurysm. After drainage and débridement of the mediastinum, replacement of the damaged segment of aorta with synthetic graft (protected by viable tissue of the greater omentum) has been proposed as the most acceptable solution.^1,5 Recent studies of both aortoesophageal and aortoenteric fistulae^6,7 have documented the superiority of cryopreserved allografts as conduits for the replacement of the affected aortic segment, because they are more resistant to infection. However, such grafts are not always available in the right shape and size, particularly in emergency cases; also, they can degenerate consequent to an immune response, and they carry an overall complication rate as high as 23%.⁸ Stent-grafting of the aorta, although proposed as an alternative, increases the risks of septic complications and is indicated only for high-risk patients who cannot tolerate open repair.⁹ Recently, Topel and colleagues¹⁰ reported a combined technique of initial stent-grafting for control of bleeding, followed by aortic segment replacement with a homograft, then surgical repair of the esophageal defect in a 2nd operation.

In our patient, the presence of a highly contaminated field led us to avoid extracorporeal circulation and in situ reconstruction. Therefore, we proceeded to an extra-anatomic aorto–aortic bypass. One could postulate that this approach might not reduce the risks of late graft infection, septic anastomotic complication, and aortic stump blowout in a potentially infected field. Indeed, such risks still exist, but it is difficult to define them, because the number of reported AEF cases is small and, to the best of our knowledge, a thoracic reconstruction similar to ours has never before been reported. However, in the management of infected abdominal aortic grafts via extra-anatomic bypass, aortic stump blowout has been reported in up to 23% of cases and reinfection in up to 27% of cases.^11,12 We believe that the chief preventive measures necessary to minimize these risks are, in order of importance, débridement of all infected tissues, followed by irrigation of the infected field with povidone-iodine solution; lifelong antibiotic maintenance of the patient; and efficient and safe management of the esophageal rupture, which in our opinion could include cervical esophagostomy, esophageal exclusion, and installation of a feeding tube via a gastrostomy (Fig. 4).

Surgical treatment of the esophageal rupture is controversial. Primary suture of the esophageal defect, although reported in cases with minimal mediastinal contamination, is associated with an increased risk of dehiscence and fistula recurrence.¹³ Primary esophagogastric anastomosis after subtotal esophagectomy may be another alternative, but it increases the rate of perioperative death. Subtotal esophageal resection, followed by gastric or colonic interposition in a later surgical procedure, could significantly minimize the risks of graft infection and seems to be the most promising solution in these patients.^5,14 In our patient, we decided to proceed to bipolar esophageal exclusion without esophageal resection. This is a well-established technique for the management of esophageal fistulae¹⁵ and may be a valid alternative to subtotal esophageal resection in cases of AEF. Esophageal exclusion enables healing and closure of the esophageal defect and can well prevent the accumulation of esophageal secretions and stomach contents in the territory of the aortic repair, thereby reducing the risk of empyema formation and septic complications.¹⁶ For feeding, we believe that a transthoracic, transdiaphragmatic feeding gastrostomy is the best solution, for it both avoids further opening of the abdomen for a separate gastrostomy or jejunostomy and minimizes damage to the stomach (that is, the gastrostomy will subsequently be used to achieve patency of the upper gastrointestinal tract). In our case, we were able to perform an uncomplicated retrosternal gastric bypass on a patient who had been well supported nutritionally. As an alternative treatment of AEF, one could argue in behalf of stent-grafting of the aorta and stent placement in the esophagus as an initial step, followed by thoracotomy to drain the empyema. However, there are no reports to support this approach. We believe that our patient's esophageal defect would not have healed by esophageal stenting and that further septic complications would have followed.

In conclusion, AEF is a potentially fatal complication of thoracic aortic aneurysm. Prompt surgical intervention is crucial in order to manage both the aneurysm and the esophageal erosion.

Acknowledgment

We would like to acknowledge the help offered by our colleague Vassilis Leivaditis, MD, for his excellent line drawings that present our operation in a clear and detailed fashion. Dr. Leivaditis is a postgraduate student in the Department of Cardiothoracic Surgery at Patras University.