Ann Thorac Surg 2001;72:1783-1788
© 2001 The Society of Thoracic Surgeons
Review
Gastroaortic fistula as an early complication of esophagectomy
Carlos Molina-Navarro, FRCSEd*a,
Shorland W. Hosking, MDa,
Stephen J. Hayward, FRCRb,
Alistair D.S. Flowerdew, MSc
a Department of General Surgery, Poole General Hospital, Poole, United Kingdom
b Department of Radiology, Royal United Hospital, Bath, United Kingdom
c Department of General Surgery, Dorset County Hospital, Dorchester, United Kingdom
* Address reprint requests to Mr Molina-Navarro, c/o Salvador Allende no 21, 29620 Torremolinos (Malaga), Spain
e-mail: molinacarlos{at}worldonline.es
 |
Abstract
|
|---|
Gastroaortic fistula following esophagogastrectomy is an uncommon and invariably fatal complication without urgent surgical intervention. We report 1 such case and review the world literature identifying 22 previous cases. It characteristically presents 2 to 3 weeks after esophago-gastrectomy with an initial herald bleed, followed by a latent period with a mean duration of 10 hours (range: 30 minutes to 3 days) and final exsanguination. Only 1 patient in this series survived. Awareness is necessary to allow prompt diagnosis and treatment of this almost invariably lethal but curable condition.
|
Introduction
|
|---|
Gastro-aortic fistula (GAF) is an uncommon early complication of esophagogastrectomy (EG). GAF usually presents 2 to 3 weeks after EG with a herald bleed, followed by a latent period and final exsanguination. The latent period has an average duration of 10 hours (30 minutes to 3 days) but because the condition is not suspected, diagnosis is usually made at postmortem examination. Of 2 patients who have undergone thoracotomy and surgical repair of a fistula, only 1 survived the procedure. Without surgical intervention, GAF is always fatal. Awareness is necessary to allow prompt diagnosis and treatment of this almost invariably lethal but curable condition. We report 1 case and review 22 previously reported cases in the world literature (Table 1).
|
Case report
|
|---|
A 44-year-old woman underwent transhiatal EG for a circumferential adenocarcinoma at the esophagogastric junction. A routine contrast swallow 1 week after the operation demonstrated a small and localized asymptomatic anastomotic leak. This was treated conservatively by delaying oral intake for 2 weeks, during which time the patient was fed through a percutaneous jejunostomy.
The patient was discharged on the 16th postoperative day and readmitted on the 21st postoperative day with a 1-hour history of hematemesis and passing blood clots per rectum. Two and a half hours after the initial bleed, she had a further massive hematemesis and collapsed. Resuscitation was attempted with large volumes of colloid and uncrossed matched blood. A Sengstaken- Blakemore balloon was inserted in an unsuccessful attempt to control massive upper gastrointestinal (GI) hemorrhage. She was transferred to the operating theater but had unrecordable blood pressure and arrested on arrival. Further transfusion and direct cardiac massage failed to obtain any positive response.
A postmortem examination revealed a 1.5-cm diameter benign gastric ulcer on the posterior wall of the upper mid-part of the stomach remnant. In its base was a 2-mm diameter fistula between the stomach and the aorta. The esophagogastric anastomosis was intact, and this ulcer bore no relation to it. There was no evidence of residual neoplastic disease.
|
Literature search method and results
|
|---|
We performed an unrestricted MEDLINE search of relevant Mesh headings and text words, and manually reviewed the bibliographic citations identifying 22 reported cases of fistulae between the thoracic gastric tube and aorta following EG. Most of the cases of GAF following EG were described in the 1950s and 1960s, and several reports included multiple cases [1–4]. None of these multiple-case papers, which were published before the introduction of the computerized MEDLINE search facility in 1966, have been quoted in the three most recent case reports [5, 6].
The first case was reported by Merendino and Emerson [1]. On the 12th postoperative day, the patient complained of chest pain and immediately had a small hematemesis; 2 hours later, the patient had a massive hematemesis and died. Autopsy revealed a fistula between the aorta and the esophagogastric anastomosis, thought to be related to a suture accidentally placed through the aortic wall.
Their second case was a 50-year-old man who underwent EG for a middle-third esophageal tumor. In this case, the gastrosplenic ligament was interposed between the aorta and the anastomosis with the specific purpose of protecting against the formation of GAF. Anastomotic leak and empyema complicated his postoperative course. He was readmitted 11 weeks postoperatively with hematemesis and died 15 hours later from torrential hemorrhage.
In the case described by Brookes and Stafford [7], the fistula was related to an ulcer at the site where the stomach had been anchored to the mediastinum. A sinus was also found between the fistula and the anastomosis.
Lookman [2] describes three cases, in which the fistula was related to the anastomosis, having no involvement with the anchoring sutures.
Le Roux [3] described 8 cases of GAF following EG from a series of 418 patients treated at the University of Edinburgh. GAF represented over 6% of the total operative mortality and, during a 10-year period, they expected to see 1 such case every 52 operated patients. In 7 out of the 8 cases, the GAF originated at the anastomosis; in the remaining case, it originated at the gastric closure line. Of those cases in which the anastomosis communicated with the aorta, it did so via an abscess cavity in 2 cases, via a fibrous track in 2 other cases, and directly into the aorta in the remaining 3 cases.
Maillard and associates [4] described 4 cases, 1 of them being the only recorded survival. All of these cases were very similar, presenting 9 to 25 days after a left thoraco-abdominal esophagectomy. The duration of the latent period ranged from "a few hours" to 12 hours. In all of the patients, the fistula was related to the anastomosis.
The cases described by Deutsch and Reis [5], Lau [6], and Rasuli and associates [8] showed benign gastric ulcers fistulating into the aorta at a distance from the anastomosis, similar to our patient.
|
Clinical syndrome
|
|---|
Most of the cases (82%) presented between the 2nd and the 6th postoperative weeks, and of these, 60% presented between days 9 and 21 postoperatively. Five of the cases presented late at 39, 56, and 80 days, and at 7 and 14 months.
An initial hemorrhage of short duration causes bright red hematemesis accompanied by syncope or hypotension (herald or sentinel bleed). This is inevitably followed in hours or even days by massive exsanguination and death. Low-grade pyrexia was present in about two-thirds of the patients, and chest pain was recorded in a little less than one-third of the cases. The latent period has an average duration of 10 hours (30 minutes to 3 days), and only 1 case presented dead on arrival to hospital [6].
This clinical picture is very similar to that of other intrathoracic enterovascular fistulae, in particular aorto-esophageal fistula (AOF) [9–15], first described by Chiari-Strassburgh in 1914 as a symptomatological triad which included: midthoracic pain, sentinel arterial hemorrhage, and final exsanguination after a symptom-free interval [16].
|
Etiology
|
|---|
It is difficult to explain why after the initial multiple-case reports of the pre-MEDLINE era, this serious complication has not been recorded in any major esophagectomy series [17–23], or in any other relevant review published in the Western World [24, 25].
The EG technique used by Le Roux included stripping of the aortic sheath, and he concluded that this predisposed to aortic erosion. The extensive dissection and the sepsis secondary to anastomotic leak were considered as the main etiological factors of GAF. The advent of prophylactic antibiotics and a less radical surgical technique may explain a reduction in incidence, but antibiotics will not completely avoid localized sepsis in the case of leaks, and the incidence of these is still reported as 0% to 10% for thoracic anastomosis and 10% to 25% for cervical anastomosis [23].
Peptic ulceration could also be responsible for this complication, particularly in those cases in which the fistula is not directly related to the anastomosis and in whom there is a longer latent period. There are numerous reports of gastric ulcers penetrating different intrathoracic organs, including the aorta [26–36]. These commonly originate from the thoracic portion of a herniated stomach. West and associates [37] compiled 95 examples of peptic ulcers penetrating the pericardium and heart.
Adhesions, disturbed gastric motility and innervation, deficient vascularization, radiation therapy, antiinflammatory medication, and even mechanical trauma from continuous cardiac and aortic pulsations have also been put forward as explanations for the development of these ulcers [8, 37], but a definite etiopathological mechanism remains unexplained.
|
Diagnosis
|
|---|
Patients presenting 2 to 6 weeks postesophagectomy with acute onset of hematemesis, with or without chest pain, should be suspected of having an aortogastric fistula and investigated urgently.
Upper GI endoscopy
Ideally, upper GI endoscopy should be performed as soon as possible, although it must be noted that only 1 patient in this series underwent such investigation and, in this case, the severity of the bleeding precluded the visualization of the source [8]. Even if the source of bleeding cannot be visualized, the presence of a large amount of fresh blood in the intrathoracic stomach a few days after an EG should be sufficient to justify an immediate thoracotomy. We do not know of any other pathology that could account for such a finding [24, 25]. Because most of the patients have a prolonged latent period, we believe that endoscopy might be valuable and we can probably use the experience learned with other aorto-enteric fistulae.
There is some experience in the endoscopy of aorto-esophageal and aortoduodenal fistulae. Khawaja and Varindani published the first endoscopic photographs of an aortoesophageal fistula (AEF), and reviewed the previous literature [14]. The aspect of AEF appears similar to that of aortoduodenal fistula (ADF), and the ADF is known to change its appearance depending on the timing of the endoscopy. In the immediate posthemorrhagic period, a pulsating submucosal mass with adherent clots is typical. The adjacent mucosa may appear bluish or gray. This finding is characteristic of extensive submucosal dissection by blood or a large intramural hematoma formation [15, 38]. The appearance may also resemble that of a fungating tumor covered in blood and, in this case, fatal bleeding can be provoked if biopsy is attempted [9].
Contrast radiology
Contrast investigations have proved disappointing and have failed to produce a diagnosis mainly because of the lack of awareness of this condition [8, 9, 39–41]. In the only case of GAF where angiography was performed, it was done with the intention of identifying and embolizing a suspected bleeding vessel. In this particular case, elective arteriography of the coeliac axis demonstrated a normal right gastroepiploic artery as the main supply to the stomach, and this was intact. There was a transient oval vascular blush projected over the 11th intercostal space posteriorly which cleared while the gastroepiploic artery was still opacified, but at the time it was not thought to be related to the stomach. There was also a curvilinear collection of contrast material in the stomach, along the greater curvature, which corresponded to the "pseudo vein" described by Ring and associates [40] as a sign of arterial hemorrhage. The patient died of exsanguination 4 hours after arteriography. The authors of this report concluded that, in this group of patients, angiographic examination should include biplane thoraco-abdominal aortography. This should, in theory, identify the fistula if it is actively bleeding [41]. Unfortunately, the experience with AEF is that the tract tends to be occluded by a blood clot during the latent period and therefore is not opacified by contrast medium. This probably also applies to GAF.
CT and MRI
There is very limited experience in the use of computed tomography (CT) and magnetic resonance imaging (MRI) scans in the diagnosis of thoracic aortoenteric fistulas [42–44]. The improved speed and reconstruction ability of spiral CT machines enables patients to be examined without undo delay. Contrast extravasation into the mediastinum may be readily demonstrated by this technique (Fig 1). Magnetic resonance provides anatomic and dynamic information but is not always suitable for imaging critically ill patients [44].
View larger version (65K):
[in this window]
[in a new window]
|
Fig 1. (A) Spiral computed tomographic scan (CT) scan, at the level of the aortic root, showing intravascular contrast medium extravasating into the paraesophageal tissues (black arrow) from a ruptured thoracic aortic aneurysm (open arrow). (B) Spiral CT scan through both lobes of the liver showing a collection of contrast (arrow) in the paraesophageal tissues. There are bilateral pleural fluid collections.
|
|
|
Treatment
|
|---|
In the 2 reported cases in which the condition was suspected, both had a successful surgical closure of the aortic defect, but only 1 patient survived. Le Roux’s patient, a 49-year-old man, presented with hematemesis 15 days after an Ivor-Lewis EG. Preparations were made for thoracotomy but, while awaiting anesthesia, he had a massive second hematemesis, 8 hours after the initial hemorrhage. He underwent immediate thoracotomy and, after transfusion and direct cardiac massage, the site of bleeding was identified and controlled. Unfortunately, after another cardiac arrest, circulation could not be restored [3].
Maillard and associates’ case presented with hematemesis 25 days after a left thoracoabdominal EG. A chest roentgenogram revealed a distended, fluid-filled stomach. Two hours after the initial bleed, he underwent a left thoracotomy with primary closure of a 2-mm aortic defect, cervical esophagostomy, and gastrotomy. He survived this operation and, a few months later, underwent successful reconstructive esophageal surgery [4].
Surgery is the only realistic therapeutic option when dealing with an enterovascular fistula, and there is no shortage of patients that have had successful operations [4, 22, 37, 45–54]. As suggested by Le Roux, we recommend a left thoracotomy approach since it allows early control of the aorta. The presence of a contaminated environment should not be an absolute contraindication to the use of prosthetic material for the repair of the aorta if primary closure is not possible [52, 54]. If the diagnosis is made during arteriography and active bleeding is occurring, it may be possible to use an intraaortic balloon catheter to temporarily occlude the fistula. Another option would be to use an intraluminal aortic stent, but this would involve introducing permanent prosthetic material into a potentially contaminated field.
|
Comment
|
|---|
This study has obvious limitations because few cases have been reported and most of them occurred over 40 years ago. There is little practical experience in the diagnosis and treatment of the condition so that some of the advice we give is based on parallelisms with other aortoenteric fistulas. However, we believe that valuable lessons can be learned.
GA fistula does not need to be necessarily fatal; there is, almost in every case, a precious latent period with a mean duration of 10 hours. Chest pain is not as constant as in the case of AEF, being reported in only 30% of cases. Low-grade pyrexia is a much more common feature, present in 60% of cases. We have found no other conditions that will produce similar symptoms, so that if significant hematemesis occurs in the early postoperative period following esophageal surgery, GA fistula should be put top of the differential diagnosis list. Diagnosis is possible with modern imaging techniques but, if strong clinical suspicion exists, surgery should not be delayed as it represents the only hope of survival.
|
Acknowledgments
|
|---|
The authors thank Mrs Irene Martos for her assistance with the translations, the librarians at Dorset County Hospital and Poole Hospital for their invaluable help, and Dr Allen M. Anscombe for allowing us to use clinical information regarding his patient.
|
References
|
|---|
-
Merendino K.A., Emerson E.C. Aortoesophagogastric fistula, an unusual complication of esophagogastrostomy performed under the aortic arch following esophageal resection for carcinoma. J Thorac Surg 1950;19:405-411.
-
Lookman A.A. Aortogastric fistula. Br J Surg 1959;46:652-655.
-
Le Roux B.T. Aortic erosion complicating oesophago-gastrectomy. Br J Surg 1961;49:271-277.
-
Maillard J.N., Blanc L., Brucher C. Aorto-esophago-gastric fistula following resection of an esophageal cancer. Description of an operative cure. Ann Clin Thorac Cardiovasc 1967;6:103-109.
-
Deutsch A.A., Reis R. Aortogastric fistula: an unusual complication of the thoracic portion of the stomach. Arch Surg 1978;113:537.[Medline]
-
Lau O.J. Acute aortogastric fistula following gastroesophageal anastomosis. Br J Surg 1983;80:504.
-
Brookes V.S., Stafford J.L. Peptic ulceration and perforation of the stomach after oesophagectomy. Thorax 1952;7:167-169.
-
Rasuli P., Hammond I., Eidus L.B., Pattee P.L. Peptic ulcer induced gastroaortic fistula: difficulty in making the diagnosis with selective angiography. J Can Assoc Radiol 1990;41:151-152.
-
Baron R.L., Koehler R.E., Gutierrez F.R., Forrest J.V., Weyman P.J. Clinical and radiographic manifestations of aortoesophageal fistulas. Radiology 1981;141:599-605.[Abstract/Free Full Text]
-
Carter R., Mulder G.A., Snyder E.N., Brewer L.A. Aortoesophageal fistula. Am J Surg 1978;136:26-30.[Medline]
-
Cronen P., Snow N., Nightingale D. Aortoesophageal fistula secondary to reflux esophagitis. Ann Thorac Surg 1982;33:78-80.[Medline]
-
Gelzayd E.A., Braucher R.E., Kirsner J.B. Aortic perforation complicating carcinoma of the esophagus: a case report and review of the literature. Gastrointest Endosc 1967;14:36-39.[Medline]
-
Grey T.C., Mittleman R.E., Wetli C.V., Horowitz S. Aortoesophageal fistula and sudden death. A report of two cases and review of the literature. Am J Forensic Med Pathol 1988;9:19-22.[Medline]
-
Khawaja F.I., Varindani M.K. Aortoesophageal fistula: review of clinical, radiographic and endoscopic features. J Clin Gastroenterol 1987;9:342-344.[Medline]
-
Sinar D.R., DeMaria A., Kataria Y.P., Thomas F.B. Aortic aneurysm eroding the esophagus. Case report and review. Am J Dig Dis 1977;22:252-254.[Medline]
-
Chiari-Strassburgh H. Injury of the esophagus with perforation of the aorta produced by a foreign body [Ueber Fremdkorperverletzung des Oesophagus mit Aortenperforation]. Berl Klin Wochenschr 1914;51:7-9.
-
Giuli R., Cignoux M. Treatment of carcinoma of the esophagus. Retrospective study of 2400 patients. Ann Surg 1980;192:44-52.[Medline]
-
Goldminc M., Maddern G., Le Prise E., Meunier B., Campion J.P., Launois B. Oesophagectomy by a transhiatal approach or thoracotomy: a prospective randomized trial. Br J Surg 1993;80:367-370.[Medline]
-
Griffin S.M., Woods S.D.S., Chan A., Chung S.C.S., Li A.K.C. Early and late surgical complications of subtotal oesophagectomy for squamous carcinoma of the oesophagus. J R Coll Surg Edinb 1991;36:170-173.[Medline]
-
Hortsmann O., Verreet P.R., Becker H., Ohmann C., Roher H.D. Transhiatal oesophagectomy compared with transthoracic resection and systematic lymphadenectomy for the treatment of oesophageal cancer. Eur J Surg 1995;161:557-567.[Medline]
-
O’Rourke I., Tait N., Bull C., Gebski V., Holland M., Johnson D.C. Oesophageal cancer: outcome of modern surgical management. Aust N Z J Surg 1995;65:11-16.[Medline]
-
Schattenkerk E.M., Obertop H., Mud H.J., Eijkenboom W.M.H., van Andel J.G., van Houten H. Survival after resection for carcinoma of the oesophagus. Br J Surg 1987;74:165-168.[Medline]
-
Urschel J.D. Esophagogastrostomy anastomotic leaks complicating esophagectomy: a review. Am J Surg 1995;169:634-640.[Medline]
-
Katz P.O., Salas L. Less frequent causes of upper gastrointestinal bleeding. Gastroenterol Clin North Am 1993;22:875-889.[Medline]
-
Rollhauser C., Fleischer D.E. Nonvariceal upper gastrointestinal bleeding: an update. Endoscopy 1997;29:91-105.[Medline]
-
Kielhofner M.A., Schnell G., Schubert T.T., Kebede-Daniels D. Aortogastric fistula from hiatal hernia ulcer. J Clin Gastroenterol 1987;9:697-700.[Medline]
-
Macarthur A.M., Wright J.E.C. Complications of intrathoracic gastric ulcer associated with hiatus hernia: a description of five cases. Br J Surg 1969;56:161-164.[Medline]
-
McDermott M., Hourihane D.O. Fatal nonmalignant ulceration in the gastric tube after oesophagectomy. J Clin Pathol 1993;46:483-485.[Abstract/Free Full Text]
-
McDonald C.F., Walbaum P.R., Sircus W., Grant I.W.B. Intrapleural perforation of peptic ulcer in association with diaphragmatic hernia. Br J Dis Chest 1985;79:196-199.[Medline]
-
O’Reilly K., Sippe R. Exsanguinating haemorrhage from the thoracic aorta following penetration by a gastric ulcer. Med J Aust 1970;2:25-26.[Medline]
-
Peters J.L., Fisher C., Kenning B.R., Pye J.K. Late benign intrathoracic gastric perforation after oesophagectomy for carcinoma. Br Med J 1981;282:1512-1513.
-
Sabety A.M., Gerard F.P. Gastroaortic fistula. Report of a case. Dis Chest 1968;53:783-784.
-
Salo J.A., Heikkila L., Nemlander A., Lindahl H., Louhimo I., Mattila S. Barrett’s esophagus and perforation of gastric tube ulceration into the pericardium: a late complication after reconstruction of esophageal atresia. Ann Chir Gynecol 1995;84:92-94.[Medline]
-
Sternberg A., Nava H.R., Isac A.T., Douglass H.O. Perforation of a benign gastric ulcer into the subdiaphragmatic aorta. Am J Gastroenterol 1987;82:579-581.[Medline]
-
Uchida Y. Occurrence of peptic ulceration in the gastric tube used for esophageal replacement in adults. Jpn J Surg 1987;17:190-194.[Medline]
-
Wasvary H., Wease G., Bierema T., Glover J. Gastroaortic fistula: an uncommon complication of Nissen fundoplication. Am Surg 1997;63:455-458.[Medline]
-
West A.B., Nolan N., O’Brian D.S. Benign peptic ulcers penetrating pericardium and heart: clinicopathological features and factors favoring survival. Gastroenterology 1988;94:1478-1487.[Medline]
-
Puppala A.R., Munaswamy M., Doshi A.M., Steinheber F.U. Endoscopic diagnosis of aortoduodenal fistula. Am J Gastroenterol 1980;73:414-417.[Medline]
-
Tonti R., Pavlica P., Ramini R., Bacchini P. Aorto-esophageal fistula [Fistola aorto-esofagea]. Radiol Med 1990;79:630-632.
-
Ring E.J., Athanasoulis C.A., Waltman A.C., Baum S. The pseudo-vein: an angiographic appearance of arterial hemorrhage. J Can Assoc Radiol 1973;24:242-244.[Medline]
-
Gregson R., Craig O. Aorto-enteric fistulae: the role of radiology. Clin Radiol 1983;34:65-72.[Medline]
-
Gutman H., Russo I., Neuman-Levin M., Haddad M., Zelikovski A. Computed tomography diagnosis of primary aorto-enteric fistula. Clin Imaging 1989;13:215-216.[Medline]
-
Kukora J.S., Rushton F.W., Cranston P.E. New computed tomographic signs of aortoenteric fistula. Arch Surg 1984;119:1073-1075.[Abstract]
-
Nagao Y., Okanoue T., Kono I., Kashima K., Kadowaki S. Magnetic resonance imaging for the diagnosis of esophago-aneurysmal fistula. Intern Med 1992;31:400-403.[Medline]
-
Agrez M.V., Ferguson N.W. Aorto gastric fistula following Nissen fundoplication. Aust N Z J Surg 1990;60:823-825.[Medline]
-
Ctercteko M.B., Mok C.K. Aorto-esophageal fistula induced by a foreign body. J Thorac Cardiovasc Surg 1980;80:233-235.[Abstract]
-
Finelli F.C., Piasio M., Mahfood S., Keshishian J.M., Golocovsky M. Esophageal rupture complicated by a gastrocardiac fistula: review of the literature and report of a successfully treated case. Ann Thorac Surg 1989;48:582-583.[Abstract]
-
Jackman J.S.G., McDonald A.M. Haematemesis due to an aorto-gastric fistula secondary to benign gastric ulceration. JR Coll Surg Edim 1987;32:377.
-
Luketich J.D., Sommers K.E., Bartley P.G., et al. Successful management of secondary aortoesophageal fistula. Ann Thorac Surg 1996;62:1852-1854.[Abstract/Free Full Text]
-
Marmuse J.P., Sevin F., Rcheid H.A. Massive digestive hemorrhage caused by aorto-esophageal fistula. J Chir (Paris) 1986;123:83-90.[Medline]
-
Schiessel R., Rath T., Kretschmer G. Giant ulcer of the oesophagus with erosion of the aorta. Successful treatment with aortic suture secured by an omental graft and esophagectomy. Br J Surg 1985;72:650.[Medline]
-
Snyder D.M., Crawford E.S. Successful treatment of primary aorta-esophageal fistula resulting from aortic aneurysm. Thorac Cardiovasc Surg 1983;85:457-463.
-
Von Oppell U.O., Groot M., Thierfelder C., Zilla P., Odell J.A. Successful management of aortoesophageal fistula due to thoracic aortic aneurysm. Ann Thorac Surg 1991;52:1168-1170.[Abstract]
-
Da Silva E.S., Tozzi F.L., Otochi J.P., de Tolosa E.M., Neves C.R., Fortes F. Aortoesophageal fistula caused by aneurysm of the thoracic aorta: successful surgical treatment, case report, and literature review. J Vasc Surg 1999;30:1150-1157.[Medline]
Top
Abstract
Introduction
