Aortoesophageal fistula: Report of a successfully managed case and review of the literature William M. Bogey, Jr., M D , James H . T h o m a s , M D , and Arlo S. H e r m r e c k , M D , P h D , Greenville, N.C., and Kansas City, Kan. Aortoesophageal fistula is a rare, frequently fatal cause of upper gastrointestinal bleeding. Although several causes have been described, it appears that the most common cause is rupture of a thoracic aortic aneurysm into the esophagus, occurring in approximately 12% of thoracic aneurysm ruptures. Although the entity was originally described in 1818, the diagnosis has rarely been made before death, and until 1983 no one survived surgery for aneurysm-associated aortoesophageal fistulas. Since that report of two cases, only two other survivors, including this patient, have been reported. Included herein is the report of a successfully managed case, along with a discussion of the causes, clinical features, and diagnostic approach to the disease, as well as the management of both the aortic and esophageal components of the fistula. (J VASC SLVRG1992;16:90-5.)
Aortoesophageal fistula (AEF) is a rare cause o f upper gastrointestinal bleeding. Although a number o f causes have been implicated, it appears that thoracic aortic aneurysm rupture is the most common cause. 1-6 Although the entity was first reported in 1818, 7 and its characteristic clinical presentation described in 1914, 8 the diagnosis is rarely made before death. 912 Even when the correct diagnosis is made, therapy has been largely unsuccessful. N o one has survived medical management, and surgical treatment has resulted in only five successfully managed cases in the English-language literature. ~3-~6 Two o f these previous survivors had fistulas as a result o f other causes, whereas only three were aneurysm related. Herein we report the sixth case successfully treated, along with a review o f the important features o f the disease and its management. CASE REPORT
A 60-year-old white woman was admitted to an outlying hospital after a single episode of "blacking out" associated with vomiting of bright red blood. Her past medical history revealed mild essential hypertension and a history of vagotomy and antrectomy for a bleeding duodenal ulcer 15 years ago. The patient described a 2-month history of interscapFrom the Departments of Surgery, Universityof Kansas Medical Center and East Carolina University, Greenville. Reprint requests: William M. Bogey, Jr., MD, Division of Vascular Surgery,Department of Surgery, Schoolof Medicine, PCMH Room 240, Greenville,NC 27858-4354. 24/4/33990 90
ular pain radiating to her neck for which she had been treated by a chiropractor. This pain was associated with dysphagia for solid foods. After admission, the patient had no further significant bleeding. Evaluation included chest radiography, which was interpreted as normal (Fig. 1), and upper gastrointestinal endoscopy. The endoscopy revealed esophageal compression by an extrinsic mass with an active bleeding site at 25 cm. No biopsy was attempted. Clot prevented adequate visualization of the greater curvature of the stomach. Postoperative changes from her previous ulcer procedure were present. The patient was transferred to our facility for further management. On arrival, the patient was awake, alert and hemodynamically stable. A dynamic CT scan of the chest was obtained and demonstrated a thoracic aortic aneurysm (Fig. 2). The patient immediately underwent a thoracic aortogram and was taken to the operating room for repair (Fig. 3). Through a left fifth intercostal space thoracotomy, a 12 mm polytetrafluoroethylene (PTFE) shunt was placed between the proxir~al left subclavian artery and the distal thoracic aorta well below the aneurysm (Fig. 4, A). The shunt was opened and the aorta was cross-clamped proximal and distal to the aneurysm. The aneurysm was then opened and its contents evacuated. With a combination of dissection from within and outside the medial wall of the aneurysm, the site of fistulization with the esophagus was identified (Fig. 4, B). Devitalized tissue was debrided, and the esophageal perforation was closed in three layers by use of absorbable suture, taking care to achieve good mucosal approximation with the innermost layer. The aneurysmal aorta was then endarterectomized and the aorta transected at the proximal and distal aneurysm necks. The
Volume 16 Number 1 July 1992
Fig. 1. Admission chest radiograph (interpreted as normal). remaining aneurysm adventitial wall was then used to patch the esophageal repair, with use of both the medial and lateral walls to achieve a two-layered patch (Fig. 4, C and D). The chest was then irrigated copiously with antibiotic containing saline solution, and aortic continuity was reestablished with a PTFE tube graft. The graft was wrapped circumferentially with a laterally based parietal pleural flap sewn in place with absorbable sutures. The PTFE shunt, which had been clamped when the aortic repair was completed, was then removed. The chest was closed in a standard fashion, and the patient was awakened and taken to the intensive care unit. The postoperative recovery was uneventful. The patient was extubated on the first postoperative day. A gastrograffin swallow on the fifth postoperative day showed no evidence of esophageal leak, and the patient was started on a diet and advanced without difficulty. She remained on intravenous antibiotics in the hospital for 1 week, and was discharged home on the eighth postoperative day on broad-spectrum oral antibiotics. These were discontinued 1 month later, and the patient continues to do well at present, 12 months since her repair.
Aortoesophagealfistula: Case report and review 91
Fig. 2. CT scan demonstrates thoracic aortic aneurysm with impingement on esophagus (nasogastric tube in esophagus causing CT artifact).
Fig. 3. Thoracic aortogram demonstrates extent of aneurysm. There is no evidence of contrast extravasation.
DISCUSSION In 1818, Debreiul, 7 a French naval surgeon, reported a case o f aortoesophageal fistula occurring in a 28-year-old soldier after ingestion o f a beef bone fragment. The soldier developed chest pain and dysphagia several hours after the ingestion, and succumbed to exsanguinating hemorrhage 5 days later. Autopsy revealed a fistula between the esophagus and thoracic aorta. Chiari s described the "Aorto-esophageal Syndrome" in 1914 consisting o f
chest pain followed by a relatively asymptomatic "latent period," a prodromal hemorrhage, and finally fatal exsanguination. Although this sequence was described after foreign body ingestion, the clinical presentation is relatively constant independent o f the origin. 17 Rupture o f a thoracic aortic aneurysm (TAA) into the esophagus was first reported by Funk in 1918.17 Over 70 cases have since been reported) 8 Autopsy
Journal of VASCULAR SURGERY
92 Bogey, Thomas, and Hermreck
A
/ D
Fig. 4. A, Esophagus and aneurysm with PTFE shunt in place. B, The fistula as viewed from within the aneurysm. C, Esophageal repair completed, first layer of patch being sewn in place. D, Aortic continuity restored, shunt removed.
series, including the collected series of 1938 cases by Hooper, 6 suggest an incidence of AEF in 9.4% to 20.4% of ruptured TAAs. Other large series have confirmed an incidence of AEF in approximately 12% of TAA ruptures, 9,1° making this the most common cause of AEF. In spite of the significant incidence of AEF in association with TAAs, the diagnosis was first made before death in 1948.19 No surgical treatment was attempted in this case. Surgical repair was attempted in 1975 and 1981, but both patients died. 9,2° The first successful repair was reported in 1983,15 and since that report of two cases, only two more successfully managed cases, including this case, have been reported. 16
Foreign body perforation of the esophagus resuiting in AEF appears to be the second most common cause of AEF, although the true incidence is unknown. Sloop and Thompson zl were able to identify 81 cases of foreigu body induced AEF in the literature in 1967, with fish and fowl bones being the most common causative agents. Nandi and Ong 22 found only two instances of AEF in their review of 2394 patients with esophageal foreign bodies, although E1 Barbary et al.2s were able to identify five instances of AEF in their 200 cases in the pediatric age group. The first reported successful repair of AEF involved a foreign body induced fistula, reported by Taniewski in 1961. 2s Aortoesophageal fistula as a complication of esophageal or other intrathoracic cancer has been reported, although it appears to be extremely ra~re.~7 Taquino et al.26 reported an incidence of four fistulas in their 145 cases, and another large series found an incidence of four fistulas in 126 cases. 27 Iatrogenic causes of AEF have been reported in relation to a number of procedures. Fistulization is known to occur after resection of esophageal carcinoma. This is most commonly caused by breakdown of the esophageal anastomosis with localized abscess formation and erosion of the aorta, 2s'3° but may occur as a result of an errantly placed suture in an esophageal anastomosis, which is inadvertently placed through the esophagus and into the media of the aorta, s~ Errant suture placement during repair of a TAA has also been cited as a cause of AEF. 3235 Other iatrogenic causes include AEF as a complication of Celestin tube placement, patent ductus arteriosus ligation, coarctation repair, and vascular ring repair, l°,~s Yonago et al. 1° reported the first successful repair of an AEF in the English-language literature in 1969. This fistula occurred after repair of an aortic ring and was managed by primary esophageal repair in combination with aortic resection and reestablishment of aortic continuity by use of a graft placed through the right chest. Diagnosis of AEF requires a high index of suspicion and an aggressive posture toward evaluation. The diagnosis should be entertained in any patient with midthoracic pain, hematemesis, and a history suggesting TAA, foreign body ingestion, or esophageal cancer.X7With TAA, symptoms of esophageal obstruction are frequently present, and these symptoms typically precede hematemesis by several months. 15 The suggestion of a posterior mediastinal mass on chest radiography is an important finding, but the chest radiograph may not show a TAA even if present, as in this case report. 9,sz
Volume 16 Number 1
july 1992
Once the diagnosis of AEF is entertained, further evaluation should proceed immediately, as the latent interval between prodromal hemorrhage and exsanguination is unpredictable? 2 Historically, most patients have died before the diagnosis was made. 9q2 Specific radiologic tests useful in the diagnosis of AEF include contrast studies of the esophagais and aorta, and CT scanning. These studies may not identify the fistula, but often confirm esophageal or aortic lesions known to be associated with AEF. 9'1°'1s The CT scan can provide definition of the esophagus and aorta in the region of the suspected fistula and can differentiate aneurysms from solid tumors,S739 Barium swallow is more likely to reveal changes indicative of AEF than aortogram.9,~ °,~2,4° In ~the study by Baron et al.9 all patients showed narrowing, extrinsic compression, and deviation of the esophagus (usually anteriorly and to the right). The compression of the esophagus may be smooth, but more commonly an ulcerating mass can be identified. 12,41,4zNaschitz et al.4° proposed a triad suggestive of AEF consisting of upper gastrointestinal hemorrhage, filling defect on esophagram, and TAA. Aortography will reveal the presence of TAA, but is unlikely to reveal extravasation diagnostic of AEF. 9 Endoscopy is probably the most sensitive and specific diagnostic technique, but it carries the risk of inducing massive hemorrhage.14 Endoscopic biopsy has been temporally related to massive bleeding in at least two patients, 9 and is therefore contraindicated in suspected AEF.37 The endoscopic appearance of AEF is similar to aortoduodenal fistula, and is influenced by the timing of the examination.4s Early on after a prodromal bleed there is a pulsating submucosal mass appearance with adherent blood clots, occasionally with a blue-grey appearance to the surrounding mucosa caused by a submucosal hematoma, n,s7 Later, during the latent period, the fistula appears as a pulsatile mass bulging into the esophagus with eroded, ulcerated overlying mucosa covered by a fibrinous exudate suggestive of a fungating neoplasm? 2 It is this "pseudotumor" that is most likely to be biopsied by the unwary endoscopist. In spite of the risk of hemorrhage, endoscopy is recommended as the primary mode of study in suspected AEF, just as in most other causes of acute gastrointestinal hemorrhage? 7 Consideration should be given, however, to performing the endoscopy in the operating suite with the patient prepared for urgent thoracotomy should bleeding result. 17 Once the diagnosis of AEF is made, treatment should proceed without delay. A left thoracotomy
Aortoesephagealfistula: Case report and review 93
incision provides the best exposure in most fistulas. 12'17 Control of the fistula is of primary concern, and all of the successfully managed cases reported thus far illustrate this well. This control requires attention to both the aortic and esophageal components, and it is beneficial to consider these portions separately. The esophageal fistula occurs as a result of chronic pressure causing erosion and ischemic necrosis of the esophagus, with chronic inflammatory changes of the tissue arotmd the perforation, is The notoriously poor wound healing of the esophagus makes this repair the most challenging aspect of management in these patients? 6 Historically, most have advocated esophageal resection with restoration of gastrointestinal continuity in either a primary or staged fashion.2'6'ls'~ This concept was applied by Crawford and Snyder~s in their report of the first two successfully repaired AEFs caused by TALk in 1983. These authors thought that this approach provided the greatest likelihood of healing without leaks that would threaten the aortic repair. This technique is clearly advisable if the esophagus is extensively damaged, or if esophageal disease or neoplasm is the cause of the fistula? °,as Recent reports, however, including a report from Coselli and Crawford 16 in 1990, as well as our own, have advocated primary esophageal repair if the esophageal lesion is limited in size. We believe that the use of the PTFE subclavianto-descending aorta temporary shunt allowed for an unhurried approach to the repair of the esophagus and thus contributed to our success. We used the endarterectomized medial and lateral wall of the aortic aneurysm to "patch" our esophageal repair. Coselli and Crawford made a separate abdominal incision and used an omental pedicle flap to reinforce their esophageal repair and wrap the aortic repair. We agree that reinforcement of the esophageal repair is important, the choice of tissue used for this purpose is probably less important. Management of the aortic aneurysm is the second component of the surgical repair ofAEF. Options for treating the aneurysm include excision or exclusion of the aneurysm with proximal and distal aortic occlusion combined with extraanatomic bypass to preserve distal blood flow, or in situ aneurysm repair? s The first option was used by Yonago et al.13 but has not been widely recommended since. Previous reports have shown that in situ repair in mycotic aneurysms and in aortoduodenal and aortojejunal fistulas as a result of aneurysm can be performed safely. 4s47 Chan et al.46 reported a series of 22 patients with mycotic aneurysms that were repaired with in situ graft
94
Journal of VASCULAR SURGERY
Bogey, Thomas, and Hermreck
replacement without the occurrence of postoperative graft infection over a follow-up period of 3 months to 8 years. They stressed the importance ofdebriding all infected tissue, and felt that lifetime antibiotic suppression was important, although four patients had their antibiotics stopped between 2 weeks and 3 years after operation without adverse sequelae. Another two patients stopped their antibiotics against medical advice, however, and both of these patients developed recurrent infections. Walker et al.45 reported a similar number of patients with secondary aortoduodenal and aortojejunal fistulas repaired with in situ grafting after excision of the involved segments of their original grafts. Eighty-three percent of their early survivors continued to do well at a mean follow-up of 5.2 years. The development of recurrent infection in their series appeared to be related to the presence of frank abscess at the time of their initial presentation and the failure to use an omental interposition as a part of the repair. They chose antibiotic treatment for periods ranging from 4 days to several weeks, although a "prolonged course" is recommended in the discussion. We would agree that the more extensive procedure required for extraanatomic bypass does not appear justified based on these reports as well as the report by Coselli and Crawford in 1990.16 Every effort should be made, of course, to be sure that the bed into which the graft is placed is as clean as possible. To this end we believe that thorough resection of all devitalized tissue and copious irrigation of the chest after esophageal repair are important. Equally important, we believe, is placement of the aortic graft external to the aneurysm to avoid proximity of the graft and the potentially unsterile inner wall of the aneurysm. We also chose to wrap our graft in a parietal pleural flap to further protect it from potential contamination. The duration of antibiotic coverage has ranged from several days to indefinitely among the previous reports of in situ graft replacement in contaminated fields without strong evidence to support any regimen chosen. The optimal period for antibiotic coverage in these patients is yet to be defined. In summary, AEF is an unusual cause of treatable upper gastrointestinal hemorrhage requiring a high index of suspicion and an aggressive posture toward evaluation and repair if it is to proceed successfully. Repair requires careful consideration of the esophageal perforation as well as the aortic lesion. With appropriate attention to each of these components, a successful surgical outcome can be achieved.
REFERENCES
1. McNamara JT, Pressler VM. Natural history of arteriosclerotic thoracic aortic aneurysm. Ann Thorac Surg 1978;26: 468-73. 2. Lemann II. Aneurysm of the thoracic aorta: its incidence, diagnosis and prognosis. A statistical study. Am J Med Sci 1966;152:210-22. 3. Loche B, Rea MH. Studies on aneurysms. II. Aneurysms of the aorta. JAMA 1923;81:1167-70. 4. Kempmeier RH. Saccular aneurysms of the thoracic aorta. A clinical study of 633 cases. Ann Intern Med 1938;12:624-51. 5. Brindiey P, Stembridge VA. Aneurysms of the aorta. A clinicopathologic study of 369 necropsy case. Am J Pathol 1956;32:67-82. 6. Hooper WL. Haematemesis and melaena due to rupture of a saccular aneurysm of the aorta into the esophagus. Postgrad Med J 1962;38:297-301. 7. Dubrueil. Observations sur la perforation de l'oesophage et de l'aorte thoracique par une potion d'os avale: Avec der reflexions. J Univ Sci Med 1818;9:357-63. 8. Chiari H. Ueber Fremdkorperverletzung des Oesophagus mit Aortenpefforation. Berlin Klin Wschr 1914;51:7-9. 9. Baron RL, Koehler RE, Gutierrez FR, et al. Clinical and radiographic manifestations of aortoesophageal fistula. Radiology 1981;141:599-605. 10. Carter R, Mulder GA, Snyder EN, et al. Aortoesophageal fistula. Am J Surg 1978;136:26-8. 11. Sinar DR, DeMaria A, Kataria YP, et al. Aortic aneurysm eroding the esophagus. Case report and review. Dig Dis Sci 1977;22:252-4. 12. Myers HS, Silber W. Oesophageal bleeding from aortoesophageal fistula due to aortic aneurysm. S Afr Med J 1983;63: 124-7. 13. Yonago RH, Iben AB, Mark JBD. Aortic bypass in the management of aortoesophageal fistula. Ann Thorac Surg 1969;7:235-7. 14. Ctercteko G, Mok CK. Aorta-esophageal fistula induced by a foreign body. The first recorded survival. J Thorac Cardiovasc Surg 1980;80:233-5. 15. Snyder DM, Crawford ES. Successful treatment of primary aortoesophageal fistula resulting from aortic aneurysm. J Thorac Cardiovasc Surg 1983;85:457-63. 16. Coselli JS, Crawford ES. Primary aortoesophageal fistula from aortic aneurysm: Successful surgical treatment by use of omental pedicle graft. J Vasc SUV,G 1990;12:269-77. 17. Funk EH. Aortic aneurysm with esophageal rupture. Med Clin North Am 1918;2:795-802. 18. Lui RC, Johnson FE, Horovitz JH, Cunningham JN. Aorto-esophageal fistula: case report and literature review. J VASC SURG 1987;6:379-82. 19. Baer S, Loewenberg SA. Aortic aneurysms simulating organic disease of the gastrointestinal tract. Gastroenterology 1948; 10:617-25. 20. Monro JL, Skidmore FD, Slokos CG, Radcliffe T. Intraesophageal rupture of a thoracic aortic aneurysm. J Cardiovasc Surg (Torino) 1975;16:302-7. 21. Sloop RD, Thompson JC. Aorto-esophageal fistula: report of a case and review of literature. Gastroenterology 1967;53: 768-77. 22. Nandi P, Ong GB. Foreign body in the oesophagus: review of 2394 cases. Br J Surg 1978;65:5-9.
Volume 16 Number 1 July 1992
23. El Barbary AS, Foad H, Fathi A. Oesophageat fistulae caused by swallowing foreign bodies. J Laryngol Otol 1969;83: 251-9. 24. Henry WJ, Miscall L. Aorto-esophageal fistula. J Thorac Cardiovasc Surg 1960;39:258-62. 25. Taniewski J. Hemorrhage from the aorta caused by a foreign body [in Polish]. Otolaryngol Pol 1961;15:367. 26. Taquino GJ, Joseph GF. Carcinoma of the esophagus. An analysis of 145 cases with special reference to metastases and extensions. Ann Otol 1947;56:1041-58. 27. Kaufman E. Pathology for students and practitioners. Philadelphia: Blakiston Son and CO, 1929;1-644. 28. Maguire WC, Mitchell N. Perforation of the aorta by acid gastric contents at site of gastresophagostomy. Surgery 1947;22:842-4. 29. LeRoux BT. Aortic erosion complicating oesophagogastrectomy. Br J Surg 1961;49:271-7. 30. Ullmann AS, Shier KJ, Horn RC. Aorto-esophageal fistula: an unusual complication of esophageal-gastrostomy, following resection for carcinoma of the esophagus. Can Med Assoc J 1961;27-31. 31. Merendino KA, Emerson EC. Aortoesophagogastric fistula. An unusual complication of esophagogastrostomy performed under the aortic arch following esophageal resection for carcinoma. A report of two cases. J Thorac Surg 1950;19: 405-11. 32. Wareing TJ, Merrill WH. Aortoesophageal fistula: unusual complication. South Med J 1989;82:1306-8. 33. Sturg BS, Sakzman DA, Feldman MI, et al. Aorto-esophageal fistula. Cardiovasc Res Cent Bull 1979;18:34-8. 34. Tierney Jr LM, Wall SD, Jacobs RA. Aorto-esophageal fistula after perigraft abscess with characteristic CT findings. J Clin Gastroenterol 1984;6:535-7. 35. Seymour EQ. Aortoesophageal fistula as a complication of aortic prosthetic graft. AJR 1978;131:160-1. 36. Lefkowitz M, Elas II LJ, Levine RJ. Candida infection complicating peptic esophageal ulcer. Infection in an aorticesophageal fistula. Arch Int Med 1964;113:672-5.
Aortoesophageal fistula: Case report and review 95
37. Sosnowik D, Greenberg R, Bank S, Grover LM. Aortoesophageal fistula: early and late endoscopic features. Am J Gastroenterol 1988;83:1401-4. 38. Goldwin RL, Heitzman ER, Proto AV. Computed tomography of the mediastinum. Normal anatomy and indications for the use of CT. Radiology 1977;124:235-41. 39. Baron RL, Levitt RG, Sagel SS, Stanley RJ. Computed tomography in the evaluation of mediastinal widening. Radiology 1981;138:107-13. 40. Naschitz JF, Bassan H, Lazarov N, et al. Upper gastrointesfinal bleeding, aneurismatic dilatation of the thoracic aorta and filling defect on the esophagogram: a diagnostic triad suggesting aortoesophageal fistula. Radiology 1982;22: 283-5. 41. Hans SY, Jander HP, Ho KJ. Aortoesophageal fistula. South Med J 1981;74:1260-2. 42. Baker MS, Baker BH. Aortoesophageal fistula. South Med J 1982;75:770-1. 43. Khawaja FI, Varindani MK. Aortoesophageal fistula: review of clinical, radiographic and endoscopic features. J Clin Gastroenterol 1987;9:342-4. 44. Dalton M, Newsom B, Isbell SA, Conn JH. Ruptured thoracic aneurysm with aortoesophageal fistula. J Miss State Med Assoc 1986;27:1-4. 45. Walker WE, Cooley DA, Duncan JM, Hallman GL Jr, Ott DA, Reul GJ. The management of aortoduodenal fistula by in situ replacement of the infected abdominal aortic graft. Ann Surg 1987;205:727-32. 46. Chan FY, Crawford ES, Coselli JS, Sail HJ, William TW Jr. In situ prosthetic graft replacement for mycotic aneurysms of the aorta. Ann Thorac Surg 1989;47:193-203. 47. Crawford ES, Palamara AE, Saleh SA, Roehm JOF. Aortic aneurysm. Current status of surgical treatment. Surg Clin North Am 1979;59:597-636.
Submitted July 9, 1991; accepted Sept. 25, 1991.
Aortoesophageal fistula (AEF) is a rare cause o f upper gastrointestinal bleeding. Although a number o f causes have been implicated, it appears that thoracic aortic aneurysm rupture is the most common cause. 1-6 Although the entity was first reported in 1818, 7 and its characteristic clinical presentation described in 1914, 8 the diagnosis is rarely made before death. 912 Even when the correct diagnosis is made, therapy has been largely unsuccessful. N o one has survived medical management, and surgical treatment has resulted in only five successfully managed cases in the English-language literature. ~3-~6 Two o f these previous survivors had fistulas as a result o f other causes, whereas only three were aneurysm related. Herein we report the sixth case successfully treated, along with a review o f the important features o f the disease and its management. CASE REPORT
A 60-year-old white woman was admitted to an outlying hospital after a single episode of "blacking out" associated with vomiting of bright red blood. Her past medical history revealed mild essential hypertension and a history of vagotomy and antrectomy for a bleeding duodenal ulcer 15 years ago. The patient described a 2-month history of interscapFrom the Departments of Surgery, Universityof Kansas Medical Center and East Carolina University, Greenville. Reprint requests: William M. Bogey, Jr., MD, Division of Vascular Surgery,Department of Surgery, Schoolof Medicine, PCMH Room 240, Greenville,NC 27858-4354. 24/4/33990 90
ular pain radiating to her neck for which she had been treated by a chiropractor. This pain was associated with dysphagia for solid foods. After admission, the patient had no further significant bleeding. Evaluation included chest radiography, which was interpreted as normal (Fig. 1), and upper gastrointestinal endoscopy. The endoscopy revealed esophageal compression by an extrinsic mass with an active bleeding site at 25 cm. No biopsy was attempted. Clot prevented adequate visualization of the greater curvature of the stomach. Postoperative changes from her previous ulcer procedure were present. The patient was transferred to our facility for further management. On arrival, the patient was awake, alert and hemodynamically stable. A dynamic CT scan of the chest was obtained and demonstrated a thoracic aortic aneurysm (Fig. 2). The patient immediately underwent a thoracic aortogram and was taken to the operating room for repair (Fig. 3). Through a left fifth intercostal space thoracotomy, a 12 mm polytetrafluoroethylene (PTFE) shunt was placed between the proxir~al left subclavian artery and the distal thoracic aorta well below the aneurysm (Fig. 4, A). The shunt was opened and the aorta was cross-clamped proximal and distal to the aneurysm. The aneurysm was then opened and its contents evacuated. With a combination of dissection from within and outside the medial wall of the aneurysm, the site of fistulization with the esophagus was identified (Fig. 4, B). Devitalized tissue was debrided, and the esophageal perforation was closed in three layers by use of absorbable suture, taking care to achieve good mucosal approximation with the innermost layer. The aneurysmal aorta was then endarterectomized and the aorta transected at the proximal and distal aneurysm necks. The
Volume 16 Number 1 July 1992
Fig. 1. Admission chest radiograph (interpreted as normal). remaining aneurysm adventitial wall was then used to patch the esophageal repair, with use of both the medial and lateral walls to achieve a two-layered patch (Fig. 4, C and D). The chest was then irrigated copiously with antibiotic containing saline solution, and aortic continuity was reestablished with a PTFE tube graft. The graft was wrapped circumferentially with a laterally based parietal pleural flap sewn in place with absorbable sutures. The PTFE shunt, which had been clamped when the aortic repair was completed, was then removed. The chest was closed in a standard fashion, and the patient was awakened and taken to the intensive care unit. The postoperative recovery was uneventful. The patient was extubated on the first postoperative day. A gastrograffin swallow on the fifth postoperative day showed no evidence of esophageal leak, and the patient was started on a diet and advanced without difficulty. She remained on intravenous antibiotics in the hospital for 1 week, and was discharged home on the eighth postoperative day on broad-spectrum oral antibiotics. These were discontinued 1 month later, and the patient continues to do well at present, 12 months since her repair.
Aortoesophagealfistula: Case report and review 91
Fig. 2. CT scan demonstrates thoracic aortic aneurysm with impingement on esophagus (nasogastric tube in esophagus causing CT artifact).
Fig. 3. Thoracic aortogram demonstrates extent of aneurysm. There is no evidence of contrast extravasation.
DISCUSSION In 1818, Debreiul, 7 a French naval surgeon, reported a case o f aortoesophageal fistula occurring in a 28-year-old soldier after ingestion o f a beef bone fragment. The soldier developed chest pain and dysphagia several hours after the ingestion, and succumbed to exsanguinating hemorrhage 5 days later. Autopsy revealed a fistula between the esophagus and thoracic aorta. Chiari s described the "Aorto-esophageal Syndrome" in 1914 consisting o f
chest pain followed by a relatively asymptomatic "latent period," a prodromal hemorrhage, and finally fatal exsanguination. Although this sequence was described after foreign body ingestion, the clinical presentation is relatively constant independent o f the origin. 17 Rupture o f a thoracic aortic aneurysm (TAA) into the esophagus was first reported by Funk in 1918.17 Over 70 cases have since been reported) 8 Autopsy
Journal of VASCULAR SURGERY
92 Bogey, Thomas, and Hermreck
A
/ D
Fig. 4. A, Esophagus and aneurysm with PTFE shunt in place. B, The fistula as viewed from within the aneurysm. C, Esophageal repair completed, first layer of patch being sewn in place. D, Aortic continuity restored, shunt removed.
series, including the collected series of 1938 cases by Hooper, 6 suggest an incidence of AEF in 9.4% to 20.4% of ruptured TAAs. Other large series have confirmed an incidence of AEF in approximately 12% of TAA ruptures, 9,1° making this the most common cause of AEF. In spite of the significant incidence of AEF in association with TAAs, the diagnosis was first made before death in 1948.19 No surgical treatment was attempted in this case. Surgical repair was attempted in 1975 and 1981, but both patients died. 9,2° The first successful repair was reported in 1983,15 and since that report of two cases, only two more successfully managed cases, including this case, have been reported. 16
Foreign body perforation of the esophagus resuiting in AEF appears to be the second most common cause of AEF, although the true incidence is unknown. Sloop and Thompson zl were able to identify 81 cases of foreigu body induced AEF in the literature in 1967, with fish and fowl bones being the most common causative agents. Nandi and Ong 22 found only two instances of AEF in their review of 2394 patients with esophageal foreign bodies, although E1 Barbary et al.2s were able to identify five instances of AEF in their 200 cases in the pediatric age group. The first reported successful repair of AEF involved a foreign body induced fistula, reported by Taniewski in 1961. 2s Aortoesophageal fistula as a complication of esophageal or other intrathoracic cancer has been reported, although it appears to be extremely ra~re.~7 Taquino et al.26 reported an incidence of four fistulas in their 145 cases, and another large series found an incidence of four fistulas in 126 cases. 27 Iatrogenic causes of AEF have been reported in relation to a number of procedures. Fistulization is known to occur after resection of esophageal carcinoma. This is most commonly caused by breakdown of the esophageal anastomosis with localized abscess formation and erosion of the aorta, 2s'3° but may occur as a result of an errantly placed suture in an esophageal anastomosis, which is inadvertently placed through the esophagus and into the media of the aorta, s~ Errant suture placement during repair of a TAA has also been cited as a cause of AEF. 3235 Other iatrogenic causes include AEF as a complication of Celestin tube placement, patent ductus arteriosus ligation, coarctation repair, and vascular ring repair, l°,~s Yonago et al. 1° reported the first successful repair of an AEF in the English-language literature in 1969. This fistula occurred after repair of an aortic ring and was managed by primary esophageal repair in combination with aortic resection and reestablishment of aortic continuity by use of a graft placed through the right chest. Diagnosis of AEF requires a high index of suspicion and an aggressive posture toward evaluation. The diagnosis should be entertained in any patient with midthoracic pain, hematemesis, and a history suggesting TAA, foreign body ingestion, or esophageal cancer.X7With TAA, symptoms of esophageal obstruction are frequently present, and these symptoms typically precede hematemesis by several months. 15 The suggestion of a posterior mediastinal mass on chest radiography is an important finding, but the chest radiograph may not show a TAA even if present, as in this case report. 9,sz
Volume 16 Number 1
july 1992
Once the diagnosis of AEF is entertained, further evaluation should proceed immediately, as the latent interval between prodromal hemorrhage and exsanguination is unpredictable? 2 Historically, most patients have died before the diagnosis was made. 9q2 Specific radiologic tests useful in the diagnosis of AEF include contrast studies of the esophagais and aorta, and CT scanning. These studies may not identify the fistula, but often confirm esophageal or aortic lesions known to be associated with AEF. 9'1°'1s The CT scan can provide definition of the esophagus and aorta in the region of the suspected fistula and can differentiate aneurysms from solid tumors,S739 Barium swallow is more likely to reveal changes indicative of AEF than aortogram.9,~ °,~2,4° In ~the study by Baron et al.9 all patients showed narrowing, extrinsic compression, and deviation of the esophagus (usually anteriorly and to the right). The compression of the esophagus may be smooth, but more commonly an ulcerating mass can be identified. 12,41,4zNaschitz et al.4° proposed a triad suggestive of AEF consisting of upper gastrointestinal hemorrhage, filling defect on esophagram, and TAA. Aortography will reveal the presence of TAA, but is unlikely to reveal extravasation diagnostic of AEF. 9 Endoscopy is probably the most sensitive and specific diagnostic technique, but it carries the risk of inducing massive hemorrhage.14 Endoscopic biopsy has been temporally related to massive bleeding in at least two patients, 9 and is therefore contraindicated in suspected AEF.37 The endoscopic appearance of AEF is similar to aortoduodenal fistula, and is influenced by the timing of the examination.4s Early on after a prodromal bleed there is a pulsating submucosal mass appearance with adherent blood clots, occasionally with a blue-grey appearance to the surrounding mucosa caused by a submucosal hematoma, n,s7 Later, during the latent period, the fistula appears as a pulsatile mass bulging into the esophagus with eroded, ulcerated overlying mucosa covered by a fibrinous exudate suggestive of a fungating neoplasm? 2 It is this "pseudotumor" that is most likely to be biopsied by the unwary endoscopist. In spite of the risk of hemorrhage, endoscopy is recommended as the primary mode of study in suspected AEF, just as in most other causes of acute gastrointestinal hemorrhage? 7 Consideration should be given, however, to performing the endoscopy in the operating suite with the patient prepared for urgent thoracotomy should bleeding result. 17 Once the diagnosis of AEF is made, treatment should proceed without delay. A left thoracotomy
Aortoesephagealfistula: Case report and review 93
incision provides the best exposure in most fistulas. 12'17 Control of the fistula is of primary concern, and all of the successfully managed cases reported thus far illustrate this well. This control requires attention to both the aortic and esophageal components, and it is beneficial to consider these portions separately. The esophageal fistula occurs as a result of chronic pressure causing erosion and ischemic necrosis of the esophagus, with chronic inflammatory changes of the tissue arotmd the perforation, is The notoriously poor wound healing of the esophagus makes this repair the most challenging aspect of management in these patients? 6 Historically, most have advocated esophageal resection with restoration of gastrointestinal continuity in either a primary or staged fashion.2'6'ls'~ This concept was applied by Crawford and Snyder~s in their report of the first two successfully repaired AEFs caused by TALk in 1983. These authors thought that this approach provided the greatest likelihood of healing without leaks that would threaten the aortic repair. This technique is clearly advisable if the esophagus is extensively damaged, or if esophageal disease or neoplasm is the cause of the fistula? °,as Recent reports, however, including a report from Coselli and Crawford 16 in 1990, as well as our own, have advocated primary esophageal repair if the esophageal lesion is limited in size. We believe that the use of the PTFE subclavianto-descending aorta temporary shunt allowed for an unhurried approach to the repair of the esophagus and thus contributed to our success. We used the endarterectomized medial and lateral wall of the aortic aneurysm to "patch" our esophageal repair. Coselli and Crawford made a separate abdominal incision and used an omental pedicle flap to reinforce their esophageal repair and wrap the aortic repair. We agree that reinforcement of the esophageal repair is important, the choice of tissue used for this purpose is probably less important. Management of the aortic aneurysm is the second component of the surgical repair ofAEF. Options for treating the aneurysm include excision or exclusion of the aneurysm with proximal and distal aortic occlusion combined with extraanatomic bypass to preserve distal blood flow, or in situ aneurysm repair? s The first option was used by Yonago et al.13 but has not been widely recommended since. Previous reports have shown that in situ repair in mycotic aneurysms and in aortoduodenal and aortojejunal fistulas as a result of aneurysm can be performed safely. 4s47 Chan et al.46 reported a series of 22 patients with mycotic aneurysms that were repaired with in situ graft
94
Journal of VASCULAR SURGERY
Bogey, Thomas, and Hermreck
replacement without the occurrence of postoperative graft infection over a follow-up period of 3 months to 8 years. They stressed the importance ofdebriding all infected tissue, and felt that lifetime antibiotic suppression was important, although four patients had their antibiotics stopped between 2 weeks and 3 years after operation without adverse sequelae. Another two patients stopped their antibiotics against medical advice, however, and both of these patients developed recurrent infections. Walker et al.45 reported a similar number of patients with secondary aortoduodenal and aortojejunal fistulas repaired with in situ grafting after excision of the involved segments of their original grafts. Eighty-three percent of their early survivors continued to do well at a mean follow-up of 5.2 years. The development of recurrent infection in their series appeared to be related to the presence of frank abscess at the time of their initial presentation and the failure to use an omental interposition as a part of the repair. They chose antibiotic treatment for periods ranging from 4 days to several weeks, although a "prolonged course" is recommended in the discussion. We would agree that the more extensive procedure required for extraanatomic bypass does not appear justified based on these reports as well as the report by Coselli and Crawford in 1990.16 Every effort should be made, of course, to be sure that the bed into which the graft is placed is as clean as possible. To this end we believe that thorough resection of all devitalized tissue and copious irrigation of the chest after esophageal repair are important. Equally important, we believe, is placement of the aortic graft external to the aneurysm to avoid proximity of the graft and the potentially unsterile inner wall of the aneurysm. We also chose to wrap our graft in a parietal pleural flap to further protect it from potential contamination. The duration of antibiotic coverage has ranged from several days to indefinitely among the previous reports of in situ graft replacement in contaminated fields without strong evidence to support any regimen chosen. The optimal period for antibiotic coverage in these patients is yet to be defined. In summary, AEF is an unusual cause of treatable upper gastrointestinal hemorrhage requiring a high index of suspicion and an aggressive posture toward evaluation and repair if it is to proceed successfully. Repair requires careful consideration of the esophageal perforation as well as the aortic lesion. With appropriate attention to each of these components, a successful surgical outcome can be achieved.
REFERENCES
1. McNamara JT, Pressler VM. Natural history of arteriosclerotic thoracic aortic aneurysm. Ann Thorac Surg 1978;26: 468-73. 2. Lemann II. Aneurysm of the thoracic aorta: its incidence, diagnosis and prognosis. A statistical study. Am J Med Sci 1966;152:210-22. 3. Loche B, Rea MH. Studies on aneurysms. II. Aneurysms of the aorta. JAMA 1923;81:1167-70. 4. Kempmeier RH. Saccular aneurysms of the thoracic aorta. A clinical study of 633 cases. Ann Intern Med 1938;12:624-51. 5. Brindiey P, Stembridge VA. Aneurysms of the aorta. A clinicopathologic study of 369 necropsy case. Am J Pathol 1956;32:67-82. 6. Hooper WL. Haematemesis and melaena due to rupture of a saccular aneurysm of the aorta into the esophagus. Postgrad Med J 1962;38:297-301. 7. Dubrueil. Observations sur la perforation de l'oesophage et de l'aorte thoracique par une potion d'os avale: Avec der reflexions. J Univ Sci Med 1818;9:357-63. 8. Chiari H. Ueber Fremdkorperverletzung des Oesophagus mit Aortenpefforation. Berlin Klin Wschr 1914;51:7-9. 9. Baron RL, Koehler RE, Gutierrez FR, et al. Clinical and radiographic manifestations of aortoesophageal fistula. Radiology 1981;141:599-605. 10. Carter R, Mulder GA, Snyder EN, et al. Aortoesophageal fistula. Am J Surg 1978;136:26-8. 11. Sinar DR, DeMaria A, Kataria YP, et al. Aortic aneurysm eroding the esophagus. Case report and review. Dig Dis Sci 1977;22:252-4. 12. Myers HS, Silber W. Oesophageal bleeding from aortoesophageal fistula due to aortic aneurysm. S Afr Med J 1983;63: 124-7. 13. Yonago RH, Iben AB, Mark JBD. Aortic bypass in the management of aortoesophageal fistula. Ann Thorac Surg 1969;7:235-7. 14. Ctercteko G, Mok CK. Aorta-esophageal fistula induced by a foreign body. The first recorded survival. J Thorac Cardiovasc Surg 1980;80:233-5. 15. Snyder DM, Crawford ES. Successful treatment of primary aortoesophageal fistula resulting from aortic aneurysm. J Thorac Cardiovasc Surg 1983;85:457-63. 16. Coselli JS, Crawford ES. Primary aortoesophageal fistula from aortic aneurysm: Successful surgical treatment by use of omental pedicle graft. J Vasc SUV,G 1990;12:269-77. 17. Funk EH. Aortic aneurysm with esophageal rupture. Med Clin North Am 1918;2:795-802. 18. Lui RC, Johnson FE, Horovitz JH, Cunningham JN. Aorto-esophageal fistula: case report and literature review. J VASC SURG 1987;6:379-82. 19. Baer S, Loewenberg SA. Aortic aneurysms simulating organic disease of the gastrointestinal tract. Gastroenterology 1948; 10:617-25. 20. Monro JL, Skidmore FD, Slokos CG, Radcliffe T. Intraesophageal rupture of a thoracic aortic aneurysm. J Cardiovasc Surg (Torino) 1975;16:302-7. 21. Sloop RD, Thompson JC. Aorto-esophageal fistula: report of a case and review of literature. Gastroenterology 1967;53: 768-77. 22. Nandi P, Ong GB. Foreign body in the oesophagus: review of 2394 cases. Br J Surg 1978;65:5-9.
Volume 16 Number 1 July 1992
23. El Barbary AS, Foad H, Fathi A. Oesophageat fistulae caused by swallowing foreign bodies. J Laryngol Otol 1969;83: 251-9. 24. Henry WJ, Miscall L. Aorto-esophageal fistula. J Thorac Cardiovasc Surg 1960;39:258-62. 25. Taniewski J. Hemorrhage from the aorta caused by a foreign body [in Polish]. Otolaryngol Pol 1961;15:367. 26. Taquino GJ, Joseph GF. Carcinoma of the esophagus. An analysis of 145 cases with special reference to metastases and extensions. Ann Otol 1947;56:1041-58. 27. Kaufman E. Pathology for students and practitioners. Philadelphia: Blakiston Son and CO, 1929;1-644. 28. Maguire WC, Mitchell N. Perforation of the aorta by acid gastric contents at site of gastresophagostomy. Surgery 1947;22:842-4. 29. LeRoux BT. Aortic erosion complicating oesophagogastrectomy. Br J Surg 1961;49:271-7. 30. Ullmann AS, Shier KJ, Horn RC. Aorto-esophageal fistula: an unusual complication of esophageal-gastrostomy, following resection for carcinoma of the esophagus. Can Med Assoc J 1961;27-31. 31. Merendino KA, Emerson EC. Aortoesophagogastric fistula. An unusual complication of esophagogastrostomy performed under the aortic arch following esophageal resection for carcinoma. A report of two cases. J Thorac Surg 1950;19: 405-11. 32. Wareing TJ, Merrill WH. Aortoesophageal fistula: unusual complication. South Med J 1989;82:1306-8. 33. Sturg BS, Sakzman DA, Feldman MI, et al. Aorto-esophageal fistula. Cardiovasc Res Cent Bull 1979;18:34-8. 34. Tierney Jr LM, Wall SD, Jacobs RA. Aorto-esophageal fistula after perigraft abscess with characteristic CT findings. J Clin Gastroenterol 1984;6:535-7. 35. Seymour EQ. Aortoesophageal fistula as a complication of aortic prosthetic graft. AJR 1978;131:160-1. 36. Lefkowitz M, Elas II LJ, Levine RJ. Candida infection complicating peptic esophageal ulcer. Infection in an aorticesophageal fistula. Arch Int Med 1964;113:672-5.
Aortoesophageal fistula: Case report and review 95
37. Sosnowik D, Greenberg R, Bank S, Grover LM. Aortoesophageal fistula: early and late endoscopic features. Am J Gastroenterol 1988;83:1401-4. 38. Goldwin RL, Heitzman ER, Proto AV. Computed tomography of the mediastinum. Normal anatomy and indications for the use of CT. Radiology 1977;124:235-41. 39. Baron RL, Levitt RG, Sagel SS, Stanley RJ. Computed tomography in the evaluation of mediastinal widening. Radiology 1981;138:107-13. 40. Naschitz JF, Bassan H, Lazarov N, et al. Upper gastrointesfinal bleeding, aneurismatic dilatation of the thoracic aorta and filling defect on the esophagogram: a diagnostic triad suggesting aortoesophageal fistula. Radiology 1982;22: 283-5. 41. Hans SY, Jander HP, Ho KJ. Aortoesophageal fistula. South Med J 1981;74:1260-2. 42. Baker MS, Baker BH. Aortoesophageal fistula. South Med J 1982;75:770-1. 43. Khawaja FI, Varindani MK. Aortoesophageal fistula: review of clinical, radiographic and endoscopic features. J Clin Gastroenterol 1987;9:342-4. 44. Dalton M, Newsom B, Isbell SA, Conn JH. Ruptured thoracic aneurysm with aortoesophageal fistula. J Miss State Med Assoc 1986;27:1-4. 45. Walker WE, Cooley DA, Duncan JM, Hallman GL Jr, Ott DA, Reul GJ. The management of aortoduodenal fistula by in situ replacement of the infected abdominal aortic graft. Ann Surg 1987;205:727-32. 46. Chan FY, Crawford ES, Coselli JS, Sail HJ, William TW Jr. In situ prosthetic graft replacement for mycotic aneurysms of the aorta. Ann Thorac Surg 1989;47:193-203. 47. Crawford ES, Palamara AE, Saleh SA, Roehm JOF. Aortic aneurysm. Current status of surgical treatment. Surg Clin North Am 1979;59:597-636.
Submitted July 9, 1991; accepted Sept. 25, 1991.
