World J Surg DOI 10.1007/s00268-016-3470-9
ORIGINAL SCIENTIFIC REPORT
Management of Tracheo- or Bronchoesophageal Fistula After Ivor-Lewis Esophagectomy R. Lambertz1 • A. H. Ho¨lscher1 • M. Bludau1 • J. M. Leers1 • C. Gutschow1 • W. Schro¨der1
Ó Socie´te´ Internationale de Chirurgie 2016
Abstract Background The development of tracheo- or bronchoesophageal fistula (TBF) after Ivor-Lewis esophagectomy remains to be a rare complication associated with a high mortality rate. Methods In this retrospective study, the charts of patients with TBF after esophagectomy were analyzed in terms of individual patient characteristics, esophagotracheal complications, respiratory function, management, and outcome. Results Between January 2000 and December 2014, 1204 patients underwent Ivor-Lewis esophagectomy for esophageal cancer; 13 patients (1.1 %) developed a TBF. In all 13 patients, a concomitant leakage of the intrathoracic esophagogastrostomy was evident, either prior to diagnosis of TBF (metachronous TBF) or simultaneously (synchronous TBF). TBF was predominantly located in the left main bronchus (n = 6, 46.1 %) or trachea (n = 5, 38.5 %). Management of TBF included re-thoracotomy (n = 7), interventional endoscopic (n = 10) or bronchoscopic therapy (n = 4). In the majority of patients (n = 8), management consisted of two subsequent treatment modalities. In 3 out of four patients, TBF was successfully treated by endoscopic stenting only. Five patients (38.5 %) died following a septic course with multiple organ failure. Conclusions The development of TBF after Ivor-Lewis esophagectomy is always combined with anastomotic leakage of the esophagogastrostomy. Treatment options primarily depend on the vascularization of the gastric conduit, the severity of the concomitant aspiration pneumonia, and the volume of the air leakage.
Introduction Transthoracic esophagectomy is the standard treatment for patients with esophageal carcinoma. Even in high-volume centers, this surgical procedure is associated with a considerable morbidity and mortality [1, 2]. TBF following esophagectomy is a rare complication contributing significantly to postoperative mortality [3–6]. Despite the
development of endoscopic and bronchoscopic techniques, the management of these fistulas remains difficult and controversial. This retrospective study aims to analyze the management and outcome of TBF diagnosed in a high-volume center for esophageal surgery and to propose a possible algorithm of different treatment options.
Patients and methods & W. Schro¨der [email protected] 1
Department of General, Visceral and Cancer Surgery, University of Cologne, Kerpener Str. 62, 50937 Cologne, Germany
In a period of 15 years (01/2000–12/2014), 1204 patients had Ivor-Lewis esophagectomy for esophageal cancer at the Department of General, Visceral and Cancer Surgery, University of Cologne. A retrospective review of all charts identified 13 patients who developed a TBF during the
123
World J Surg
postoperative course. TBF was defined as a connecting fistula between the neo-esophagus and the trachea or the neo-esophagus and the bronchial tree. Of all 13 patients, the postoperative period in terms of esophagotracheal complications, respiratory function, management, and outcome were analyzed. The median age of the patients was 57 years (range 29 to 72 years, mean 58.7 years). Eleven patients were male and two patients female. Six patients (46.1 %) presented with a squamous cell carcinoma and seven patients (53.9 %) with an adenocarcinoma (Table 1). Tumors were staged according to the TNM classification of the Union for International Cancer Control staging system [7]. Due to a locally advanced cancer, seven patients were treated with combined neoadjuvant radiochemotherapy and 1 patient had only neoadjuvant chemotherapy, according to a standardized protocol [8, 9]. After preoperative risk assessment, all patients underwent a transthoracic en-bloc esophagectomy with 2-field lymphadenectomy. Reconstruction was done with a gastric tube and an intrathoracic esophagogastrostomy (Ivor-Lewis esophagectomy). The surgical procedure is explained in detail elsewhere [10, 11]. Postoperative management Leakage of anastomosis was classified according to Low et al. into type I–III depending on the treatment of the leakage [2]. Type I leakage does not require any change in therapy or is treated medically, whereas type II leakage is defined by the necessity of interventional therapy (i.e., stent, EndoVac system, or drainage) and type III leakage Table 1 Characteristics of 13 patients with tracheo- or bronchoesophageal fistula after Ivor-Lewis esophagectomy No.
by the need of re-operation. Postoperative complications were graded I–V following the Dindo–Clavien Classification [12]. Pneumonia was defined using the Utrecht Scoring System for pneumonia following esophagectomy [13]. In case of a suspected leakage, an endoscopy for checkup of esophagogastric anastomosis was initiated as standard procedure. Endoscopic treatment consisted of an esophageal stent placement, either with a self-expending, covered metal stent (Ultraflex, Boston Scientific, Natick, MA, USA) or a self-expending, covered silicone stent (Polyflex, Boston Scientific). Bronchoscopy was conducted when TBF was strongly suspected (i.e., respiratory and/or cardiovascular failure, high leakage volume, preceding anastomotic leakage). Closure of TBF via bronchoscopy was realized by implantation of a self-expandable covered Y-stent (Leufen aerstent, Leufen Medical, Aachen, Germany). Re-thoracotomy was mandatory when a progressive septic course accompanied by multiple organ failure occurred or an ischemic gastric conduit was diagnosed. Depending on the intraoperative vascularization of the gastric conduit, the gastric tube was resected or re-anastomosed to the esophageal remnant. The tracheobronchial defect was covered with a pediculated pericardial or muscle flap (left sternocleidmastoideus muscle). The collected data were documented using Microsoft Excel 2010Ò. Statistical analysis was done by descriptive means only. PubMed database was searched for keywords ‘Ivor-Lewis esophagectomy,’ ‘tracheobronchial fistula,’ ‘anastomotic leakage,’ ‘gastrotracheal fistula,’ and ‘esophagobronchial fistula’ in order to compare the clinical data with previous reports.
Results
Patients and tumor characteristics Age
ASA
pTNM
Tumor entity
RTX/CTX
1
45
2
ypT0N0M0
AC
Yes
2 3
69 66
2 2
ypT1N0M0 ypT1N0M0
SCC SCC
Yes Yes
4
72
2
pT3N1M0
AC
No
5
70
3
pT1N0M0
SCC
No
6
57
2
ypT2N0M0
SCC
Yes
7
66
2
ypT3N1M0
SCC
Yes
8
68
2
pT3N0M0
AC
No
9
29
2
pT1N0M0
AC
No
10
55
2
ypT3N0M0
AC
Yes
11
54
3
pT1N0M0
AC
No
12
57
1
ypT1N0M0
AC
Yes
13
55
1
ypT3N1M0
SCC
Yes
AC adenocarcinoma, RTX/CTX radiochemotherapy, SCC squamous cell carcinoma
123
Postoperative complications In none of the 13 patients with TBF, an intraoperative injury of the tracheobronchial system was documented. All 13 patients were diagnosed to have an anastomotic leakage of the esophagogastrostomy either prior to the diagnosis of TBF (metachronous TBF, n = 6, 46.2 %) or simultaneously (synchronous TBF, n = 7, 53.8 %). Table 2 shows the clinical presentation in detail. A synchronous TBF was diagnosed by bronchoscopy in seven patients. TBF of the left main bronchus was the most frequent localization (6 patients, 46.1 %). Five patients (38.5 %) presented with a fistula into the trachea and two patients (15.4 %) into the right main bronchus. Synchronous TBF was located either in the left main bronchus (n = 4; 57.1 %) or the trachea (n = 3; 42.9 %). The size of the fistula varied from 3 to 15 mm in diameter (median
World J Surg Table 2 Clinical presentation and complications of 13 patients with tracheo- or bronchoesophageal fistula after Ivor-Lewis esophagectomy No
Esophagotracheal complications
Sepsis/MOF
Anastomotic Leakage (POD)
Type of leakage (I–III)a [2]
TBF (POD)
Localization of TBF
Size of TBF (mm)
Respiratory failure mechanical ventilation [72 h
Reintubation
Tracheostomy
Cumulative days of mechanical ventilation (days)
Cardiovascular failure
1
12
III
12
LMB
4
Yes
Yes
Yes
21
Yes
2
3
III
19
LMB
10
Yes
Yes
Yes
22
No
3
18
II
18
T
3
Yes
Yes
Yes
23
No
4
8
II
8
LMB
5
Yes
Yes
Yes
13
No
5
3
III
18
RMB
10
Yes
Yes
Yes
16
Yes
6 7
8 26
II II
37 27
LMB T
8 n/a
Yes Yes
Yes Yes
Yes Yes
94 7
Yes Yes
8
2
III
22
RMB
n/a
Yes
Yes
Yes
20
Yes
9
9
II
9
T
4
No
No
No
0,4
No
10
12
II
12
LMB
5
Yes
Yes
Yes
8
Yes
11
6/281
II
281
LMB
9
No
No
No
0,6
No
12
6
II
14
T
n/a
No
No
No
0,4
No
13
6
II
56
T
15
Yes
Yes
Yes
73
Yes
CRP C-reactive protein, LMB left main bronchus, n/a not available, PCT procalcitonin, POD postoperative day, RMB right main bronchus, T trachea, TBF tracheo- or bronchoesophageal fistula a
At the time of TBF diagnosis; at the time of first detection
6.5 mm). The majority of the patients (n = 10; 76.9 %) had to be re-intubated due to acute respiratory failure. In all ten patients, prolonged ventilation eventuated (cumulative mechanical ventilation: median 47 days, range 7–94) and a tracheostomy was performed. All patients developed a severe pneumonia in means of Utrecht pneumonia scoring C2 with pulmonary radiography findings, elevated leucozyte count (median 16.1 9 109/l, range: 8.4–33.2), and fever. With regard to the Dindo–Clavien classification, grade IIIa complications were recorded in four patients (30.8 %), grade III b and grade IVa in two patients, respectively (15.4 %). The hospital mortality (grade V complication) was 38.4 % (5 of 13 patients). Postoperative management Solitary endoscopic treatment was carried out in four patients (no. 3, 9, 10, and 12) and consisted of inserting an esophageal stent with the intention of simultaneous closure of both, anastomotic leakage and TBF (Table 3). This procedure was successful in 3 out of 4 patients (no. 3, 9, and 12). All three patients had fistulization to the trachea. Solitary endoscopic treatment of TBF was not successful in one patient (no. 10) with fistulization into the left main bronchus who suffered from a pneumonic sepsis and finally multiple organ failure. The patient died on postoperative day (POD) 20.
Four patients (no. 4, 7, 8, and 13) received endoscopic intervention with subsequent surgical re-exploration during the postoperative course. Two patients had endoscopic intervention with fibrin glue application (no. 4) or implantation of an esophageal EndoVac system (no. 7) because of anastomotic leakage with synchronous TBF. The EndoVac system describes the endoluminal placement of a vacuum-sealed sponge at the site of the esophageal anastomotic leakage. This sponge is connected to an external pump via a nasogastric tube applying a negative pressure. It has to be replaced every 3–4 days until complete healing of the anastomosis. Because of progressive respiratory failure in both cases, re-thoracotomy with resection of the gastric conduit and construction of a cervical esophagostomy was done. The bronchial defect was sutured and reinforced with a pediculated pericardial flap. Patient no. 8 underwent re-thoracotomy with re-anastomosis because of an anastomotic failure on POD 2. Due to a recurrent anastomotic leakage 18 days later the gastric conduit was finally resected and a cervical esophagostomy constructed. A concomitant laceration of the right main bronchus was closed with a primary suture. Two days later, a recurrent leakage of the right main bronchus was diagnosed and a pediculated pericardial flap was used to reinforce the bronchial suture line. The patient finally died on POD 157 following a cardiac arrest. Patient no. 13 received esophageal stent placement on POD 6 because of an
123
World J Surg Table 3 Synopsis of management and outcome of 13 patients with TBF No
Treatment of anastomotic leakage and TBF
Postoperative course
Outcome
Endoscopic yes/no (POD)
Bronchoscopic yes/no (POD)
Surgery yes/no (POD)
Procedures (POD)
ICU stay after esophagectomy (in days)
Discharge from hospital (POD)
Hospital mortality (POD)
Followup
1
No
Yes (12/13)
Yes (13)
Y-stent trachea (12); RT, RA, pPF, Y-stent trachea (13)
37
56
No
Alive
2
No
Yes (20)
Yes (3)
RT, RA (3); Y-stent trachea (20)
55
85
No
Alive
3
Yes (18)
No
No
Esophagus stent (18)
30
88
No
Alive
4
Yes (8)
No
Yes (29)
Fibrin glue (8); RT, GR, tEG, pPF (29)
55
106
No
Dead
5
No
No
RT, RA (4); RT, GR, tEG, pPF (19)
19
–
Yes (20)
Dead
6
Yes (8)
Yes (38)
No
Esophagus stent (8), Stent left main bronchus (38)
103
–
Yes (109)
Dead
7
Yes (26)
No
Yes (27)
Esophagus EndoVac (26), RT, GR, tEG, pPF (27)
57
90
No
Alive
8
Yes (20)
No
Yes (2/22/ 24)
RT,RA (2), Esophagus stent 20), RT, GR, suture of fistula (22), RT, pPF, fibrin glue (24)
131
–
Yes (130)
Dead
9
Yes (9)
No
No
Esophagus stent (9)
6
53
No
Alive
10
Yes (12)
No
No
Esophagus stent (12)
9
–
Yes (20)
Dead
11
Yes (6/281)
Yes (281)
No
Esophagus stent (6/281), Y-stent trachea (281)
20
61
No
Alive
12
Yes (6/14)
No
No
Esophagus stent (6), Replacement of Esophagus stent(14)
15
50
No
Alive
13
Yes (6)
No
Yes (56)
Esophagus stent (6), RT, RA, pSF (56)
74
–
Yes (79)
Dead
n = 10 (76.9 %)
n = 4 (30.8 %)
n=7 (53.8 %)
37 days (6–131)
73 days (50–106)
Yes n = 5 (38.5 %)
Yes (4/19)
GR resection of gastric conduit, ICU intensive care unit, POD postoperative day, pPF pediculated pericardial flap, pSF pediculated sternocleidmastoideus flap, RA re-anastomosis of esophagogastric anastomosis, RT re-thoracotomy, TBF tracheo- or bronchoesophageal fistula, tEG temporary cervical esophagostomy
anastomotic leakage. On POD 56, removal of esophageal stent revealed a TBF to the trachea. After left cervical incision with upper sternal split, the cervical defect of the trachea was primarily sutured and reinforced with the interposition of the mobilized left sternocleidmastoideus muscle. Despite this surgical intervention, the patient died on POD 79. Endoscopic esophageal stenting as well as bronchoscopic stenting (double stenting) was performed in two patients (no. 6 and 11). After initial double stenting (no. 6), a later endoscopic control of the esophageal stent revealed an enlargement of the anastomotic defect to half of the circumference as well as the TBF. The patient died without possible surgical options on POD 109. Patient no. 11 had a history of early anastomotic leakage on POD 6 which was successfully closed of an esophageal stent. After stent removal, multiple pneumatic dilatation of the stenotic
123
esophagogastrostomy had to be performed and re-insufficiency of anastomosis with a secondary TBF into the left main bronchus was visualized on POD 281. The patient received an interventional treatment with insertion of a second esophageal stent combined with a tracheal Y-stent. Two patients (no. 1 and 2) were treated by bronchoscopic intervention and surgery. In patient no. 1 who received a tracheal Y-stent for TBF repair on POD 12, a failure of mechanical ventilation due to a high-volume air leakage led to emergency surgery on the following day consisting of re-thoracotomy, re-anastomosis, and pediculated pericardial flap for fistula closure as well as reinsertion of a tracheal Y-stent. The re-operation was done using an extracorporal membranous oxygenation (ECMO). Patient no. 2 had re-thoracotomy with re-anastomosis of esophageal anastomosis because of early anastomotic leakage on POD 3. Eighteen days later, bronchoscopy
World J Surg
revealed a 10-mm-wide defect of the left main bronchus. A tracheal Y-stent was inserted, leading to a successful closure of TBF. The last patient (no. 5) of this series was treated only by surgical re-exploration. After development of early anastomotic leakage with following re-thoracotomy and reanastomosis on POD 4, bronchoscopy showed a fistula to the right main bronchus. Failure of mechanical ventilation made re-thoracotomy mandatory, and a gastrointestinal stent insertion was impossible due to ischemia of the gastric conduit. Resection of the ischemic gastric tube and a cervical esophagostomy was performed. The leakage of the right main bronchus was covered with a pediculated pericardial flap. The patient died 1 day after re-operation.
Discussion Transthoracic esophagectomy remains the best surgical option for cure of esophageal cancer and is therefore established as a standard treatment in most specialized centers for esophageal surgery [14]. Nevertheless, transthoracic esophagectomy with gastric reconstruction continues to be a challenging procedure since it is still accompanied by a considerably high rate of postoperative morbidity accounting to 40–50 % [1, 15–17]. The occurrence of TBF following transthoracic esophagectomy is rare, accounting to a reported incidence of 0.8–3.9 % [3, 5, 18, 19]. However, TBF is related with the highest mortality rate of all major complications reaching up to 57 % (Table 4). This retrospective study confirms these two basic features of TBF with an incidence of 1.1 % and a mortality rate of 38.5 %. At present, the etiology of TBF is not completely understood, but some confounding factors seem to trigger the development of TBF and support the idea of a multifactorial process. Dissection of smaller bronchial arteries may induce segmental ischemia of the tracheobronchial tree. This partial devascularization may occur after extensive peritracheal and peribronchial nodal dissection as part of the routinely performed mediastinal lymphadenectomy [3, 20]. Segmental ischemia of the vulnerable pars membranacea may be aggravated by the intraluminal pressure either by the cuff of the double-lumen tube in the left main bronchus during intraoperative single-lung ventilation or the tracheal cuff of a standard tube during prolonged mechanical ventilation postoperatively. In addition, a direct injury of the posterior aspect of the tracheobronchial tree mainly by electrocautery may contribute to a localized disturbance of microcirculation. The analysis of these retrospective data reveals another causal factor of TBF. As observed in other series [5, 18, 21], all our patients with TBF had an additional leakage of the
esophagogastrostomy. Since twelve patients had a failure of the circular and one patient of the longitudinal stapler line, it appears that not the anastomotic leakage itself but the accompanying mediastinitis with its inflammatory process next to tracheobronchial tree is the causal agent for the development of TBF [5]. The incidence of postoperative anastomotic failure is higher than that of TBF which means that not all patients with anastomotic leakage will necessarily develop a TBF. Therefore, the pathophysiological mechanism of TBF is most likely to be that of an intraoperatively induced ischemic damage of the pars membranacea with an additional postoperative inflammation of the mediastinum. In contrast to other series [3, 19], a direct injury of the trachea or the main bronchi with the need for a change of intraoperative management was not seen even in patients with SCC located at the level of tracheal bifurcation or above. This might be due to an exclusion of patients with a suspected T4 category who are at high risk of tracheobronchial injury during esophageal dissection. Therefore, it is of clinical importance that in all patients of this series the ischemic injury of the tracheobronchial system was not obvious during esophagectomy and the surgeon was generally not aware of the risk factor ‘ischemic damage.’ Several previous reports have tried to classify the localization of the TBF depending on the underlying cause. Yasuda et al. described ten patients with TBF following transthoracic esophagectomy and classified these cases into an ‘anastomotic leakage type’ developing in the trachea, the ‘gastric necrosis type’ observed predominantly around the carina, and the ‘gastric ulcer type’ associated with a fistula in the right main bronchus [6]. Bartels et al. classified TBF into an ‘ischemic’ and ‘non-ischemic’ type. The ‘ischemic’ type was localized around the carina and both main bronchi, and the ‘non-ischemic’ type was exclusively located in the trachea [3]. In this study, two predominant localizations of TBF could be identified and TBF could be classified with respect to the topographic relation to the anastomotic site into a direct (tracheal) and an indirect (left main bronchus) type. Only 2 out of 13 fistulas were located in the right main bronchus. The indirect fistulae to the left main bronchus predominantly present as metachronous TBF. Management of TBF is still difficult because this is a rare complication, and even for high-volume centers it is laborious to gain sufficient experience to modify its management and to implement a treatment algorithm. In addition, several completely different treatment strategies are under discussion all of which do not predict individual outcome at present. Therefore, treatment of TBF is still an individual decision making. However, based on the data of this retrospective study and previous reports some general recommendation for management of TBF can be made. The predominant clinical feature of TBF is acute
123
123
2015
Cologne
13
6
7
11 10
2
6
31
Patients with TBF after esophagectomy
18 (8–281)
25 (14–30)
n/a
n/a 13 (8–35)
n/a
30 (18–425)
12 (1–30)
Days to TBF median (range)
13 (100 %)
4 (66.6 %)
7 (100 %)
n/a 6 (60 %)
2 (100 %)
6 (100 %)
7 (22.5 %)
Anastomotic leakage
5 (38.5 %)
5 (83.3 %)
3 (42.8 %)
n/a 9 (90 %)
0 (0.0 %)
5 (83.3 %)
19 (61.3 %)
6 (46.1 %)
1 (16.7 %)
0 (0.0 %)
n/a 1 (10 %)
0 (0.0 %)
0 (0.0 %)
4 (12.9 %)
2 (15.4 %)
0 (0.0 %)
4 (57.2 %)
n/a 0 (0.0 %)
2 (100 %)
1 (16.7 %)
8 (25.8 %)
* Mortality rate with regard to the total number of 35 patients
0 (0.0 %)
2 (33.4 %)
1 (14.3 %)
0 (0.0 %) 4 (40 %)
0 (0.0 %)
2 (33.4 %)
n/a
Conservative treatment
RMB
Trachea
LMB
Treatment TBF
Localization of TBF
LMB left main bronchus, n/a not available, RMB right main bronchus, TBF tracheo- or bronchoesophageal fistula
2012
2010 2012
Shen et al. [4] Yasuda et al. [6]
2015
2002
Mangi et al. [21]
Morita et al. [19]
2001
Buskens et al. [18]
Schweigert et al. [5]
1998
Bartels et al. [3]
Date
Table 4 Synopsis of 86 cases of tracheobronchial fistula following Ivor-Lewis esophagectomy reported in the literature
4 (30.8 %)
0 (0.0 %)
2 (28.6 %)
0 (0.0 %) 0 (0.0 %)
0 (0.0 %)
0 (0.0 %)
n/a
Bronchoscopic treatment
7 (53.8 %)
4 (66.6 %)
2 (28.6 %)
11 (100 %) 6 (60 %)
2 (100 %)
4 (66.6 %)
n/a
Surgical treatment
5 (38.5 %)
0 (0.0 %)
4 (57.1 %)
2 (5.7 %*) 3 (30 %)
0 (0.0 %)
0 (0.0 %)
10 (32.2 %)
Hospital mortality
World J Surg
World J Surg
respiratory failure and the vast majority of patients with TBF require long-time mechanical ventilation with the need for tracheostomy. Progressive respiratory failure is primarily caused by two conditions, first the continuous aspiration through the fistula with subsequent pneumonia defined by the Utrecht score [13] and secondly an increasing air leakage which makes sufficient oxygenation difficult or even impossible. The third condition that has to be taken into consideration is the defect of the esophagogastric anastomosis and the vascularization of the gastric conduit. It is obvious that for patients with partial or complete necrosis of the gastric tube conservative (interventional) therapy is not a possible option. These patients definitely require re-thoracotomy with resection of the gastric conduit or re-do of the anastomosis. Although the re-anastomosis in the upper mediastinum or even upper thoracic inlet is a technically demanding procedure and depends on the length, the wall condition, and vascularization of the conduit, it should be the surgical therapy of choice since the gastric conduit aids to cover the tracheal or bronchial defect. In case of a conduit resection, intraoperative management of the functional air leakage is more difficult since the defect is only covered with a pediculated muscle or pericardial flap. This is particularly the case for a fistula of the left main bronchus in which the direct contact of the cuff of the double-lumen tube may enlarge the defect and suturing of any kind of flap closely to the defect is almost impossible. Patients with progressive respiratory failure due to an uncontrollable air leakage are the second patient group requiring re-thoracotomy. The principles of surgical management are similar to those patients having surgery for poor vascularization of the gastric conduit. In order to avoid this critical intraoperative condition, ventilation management includes the reduction of the airway pressure as much as possible aiming to achieve spontaneous ventilation via tracheostomy. For all other patients who do not require surgical therapy, the primary goal of treatment is the closure of the fistula in order to prevent progressive aspiration pneumonia with deterioration of respiratory function. Since the reflux from the gastrointestinal site is the etiologic agent of progressive pneumonia, stenting of the anastomotic leakage appears to be the interventional therapy of choice. This treatment option is well accepted for a solitary anastomotic leakage [22] but has also been suggested by other groups for treatment of TBF [5]. After insertion of a self-expandable stent covering the anastomotic defect, it is important to exclude persistent retrograde reflux by a CT scan with oral contrast medium applied via a nasogastric tube. As demonstrated in some patients of this series with a TBF of the trachea, treatment with a solitary gastroesophageal stent is feasible to achieve the primary goal of fistula closure. However, for these
patients with the tracheal type fistula double stenting is not indicated because the direct contact of the gastrointestinal and tracheal stent prevents granulation tissue to overgrow the existing defect. In this retrospective series, only four patients received some kind of a tracheal or bronchoscopic stent, always combined with either surgical or endoscopic treatment. Clinical management of these patients is rather difficult because the tracheobronchial stent acts as a foreign body and induces persistent pulmonary secretion. TBF is a rare but life-threatening complication with a high postoperative mortality rate. Fistulization predominantly goes to the left main bronchus or trachea and is triggered by intrathoracic leakage of the esophagogastrostomy and its inflammatory process of the mediastinum. Therefore, the treatment options primarily depend on the vascularization of the gastric conduit and the respiratory function defined by the severity of the aspiration pneumonia and the volume of the concomitant air leakage. Patients with anastomotic leakage and gastric necrosis always require immediate re-thoracotomy as well as patients with progressive respiratory failure. In all other patients, the closure of the fistula can be primarily tried to achieve from the gastrointestinal site with a self-expanding stent.
References 1. Sauvanet A, Mariette C, Thomas P et al (2005) Mortality and morbidity after resection for adenocarcinoma of the gastroesophageal junction: predictive factors. J Am Coll Surg 201:253–262 2. Low DE, Alderson D, Cecconello I et al (2015) International consensus on standardization of data collection for complications associated with esophagectomy: esophagectomy complications consensus group (ECCG). Ann Surg 262:286–294 3. Bartels HE, Stein HJ, Siewert JR (1998) Tracheobronchial lesions following oesophagectomy: prevalence, predisposing factors and outcome. Br J Surg 85:403–406 4. Shen KR, Allen MS, Cassivi SD et al (2010) Surgical management of acquired nonmalignant tracheoesophageal and bronchoesophageal fistulae. Ann Thorac Surg 90:914–918 discussion 919 5. Schweigert M, Dubecz A, Beron M et al (2012) Management of anastomotic leakage-induced tracheobronchial fistula following oesophagectomy: the role of endoscopic stent insertion. Eur J Cardiothorac Surg 41:e74–80 6. Yasuda T, Sugimura K, Yamasaki M et al (2012) Ten cases of gastro-tracheobronchial fistula: a serious complication after esophagectomy and reconstruction using posterior mediastinal gastric tube. Dis Esophagus 25:687–693 7. Wittekind C, Meyer HJ (2013) TNM: Klassifikation Maligner Tumoren, 7. Auflage 8. Cunningham D, Allum WH, Stenning SP et al (2006) Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med 355:11–20 9. van Hagen P, Hulshof MC, van Lanschot JJ et al (2012) Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med 366:2074–2084
123
World J Surg 10. Holscher AH, Schneider PM, Gutschow C et al (2007) Laparoscopic ischemic conditioning of the stomach for esophageal replacement. Ann Surg 245:241–246 11. Schroder W, Holscher AH, Bludau M et al (2010) Ivor-Lewis esophagectomy with and without laparoscopic conditioning of the gastric conduit. World J Surg 34:738–743. doi:10.1007/s00268010-0403-x 12. Dindo D, Demartines N, Clavien PA (2004) Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 240:205–213 13. van der Sluis PC, Verhage RJ, van der Horst S et al (2014) A new clinical scoring system to define pneumonia following esophagectomy for cancer. Dig Surg 31:108–116 14. Rice TW, Rusch VW, Apperson-Hansen C et al (2009) World wide esophageal cancer collaboration. Dis Esophagus 22:1–8 15. McCulloch P, Ward J, Tekkis PP (2003) Mortality and morbidity in gastro-oesophageal cancer surgery: initial results of ASCOT multicentre prospective cohort study. BMJ 327:1192–1197 16. Bailey SH, Bull DA, Harpole DH et al (2003) Outcomes after esophagectomy: a ten-year prospective cohort. Ann Thorac Surg 75:217–222 discussion 222
123
17. Kutup A, Nentwich MF, Bollschweiler E et al (2014) What should be the gold standard for the surgical component in the treatment of locally advanced esophageal cancer: transthoracic versus transhiatal esophagectomy. Ann Surg 260:1016–1022 18. Buskens CJ, Hulscher JB, Fockens P et al (2001) Benign tracheoneo-esophageal fistulas after subtotal esophagectomy. Ann Thorac Surg 72:221–224 19. Morita M, Saeki H, Okamoto T et al (2015) Tracheobronchial fistula during the perioperative period of esophagectomy for esophageal cancer. World J Surg 39:1119–1126. doi:10.1007/ s00268-015-2945-4 20. Fujita H, Kawahara H, Hidaka M et al (1988) An experimental study on viability of the devascularized trachea. Jpn J Surg 18:77–83 21. Mangi AA, Gaissert HA, Wright CD et al (2002) Benign broncho-esophageal fistula in the adult. Ann Thorac Surg 73:911–915 22. Leers JM, Vivaldi C, Schafer H et al (2009) Endoscopic therapy for esophageal perforation or anastomotic leak with a self-expandable metallic stent. Surg Endosc 23:2258–2262
ORIGINAL SCIENTIFIC REPORT
Management of Tracheo- or Bronchoesophageal Fistula After Ivor-Lewis Esophagectomy R. Lambertz1 • A. H. Ho¨lscher1 • M. Bludau1 • J. M. Leers1 • C. Gutschow1 • W. Schro¨der1
Ó Socie´te´ Internationale de Chirurgie 2016
Abstract Background The development of tracheo- or bronchoesophageal fistula (TBF) after Ivor-Lewis esophagectomy remains to be a rare complication associated with a high mortality rate. Methods In this retrospective study, the charts of patients with TBF after esophagectomy were analyzed in terms of individual patient characteristics, esophagotracheal complications, respiratory function, management, and outcome. Results Between January 2000 and December 2014, 1204 patients underwent Ivor-Lewis esophagectomy for esophageal cancer; 13 patients (1.1 %) developed a TBF. In all 13 patients, a concomitant leakage of the intrathoracic esophagogastrostomy was evident, either prior to diagnosis of TBF (metachronous TBF) or simultaneously (synchronous TBF). TBF was predominantly located in the left main bronchus (n = 6, 46.1 %) or trachea (n = 5, 38.5 %). Management of TBF included re-thoracotomy (n = 7), interventional endoscopic (n = 10) or bronchoscopic therapy (n = 4). In the majority of patients (n = 8), management consisted of two subsequent treatment modalities. In 3 out of four patients, TBF was successfully treated by endoscopic stenting only. Five patients (38.5 %) died following a septic course with multiple organ failure. Conclusions The development of TBF after Ivor-Lewis esophagectomy is always combined with anastomotic leakage of the esophagogastrostomy. Treatment options primarily depend on the vascularization of the gastric conduit, the severity of the concomitant aspiration pneumonia, and the volume of the air leakage.
Introduction Transthoracic esophagectomy is the standard treatment for patients with esophageal carcinoma. Even in high-volume centers, this surgical procedure is associated with a considerable morbidity and mortality [1, 2]. TBF following esophagectomy is a rare complication contributing significantly to postoperative mortality [3–6]. Despite the
development of endoscopic and bronchoscopic techniques, the management of these fistulas remains difficult and controversial. This retrospective study aims to analyze the management and outcome of TBF diagnosed in a high-volume center for esophageal surgery and to propose a possible algorithm of different treatment options.
Patients and methods & W. Schro¨der [email protected] 1
Department of General, Visceral and Cancer Surgery, University of Cologne, Kerpener Str. 62, 50937 Cologne, Germany
In a period of 15 years (01/2000–12/2014), 1204 patients had Ivor-Lewis esophagectomy for esophageal cancer at the Department of General, Visceral and Cancer Surgery, University of Cologne. A retrospective review of all charts identified 13 patients who developed a TBF during the
123
World J Surg
postoperative course. TBF was defined as a connecting fistula between the neo-esophagus and the trachea or the neo-esophagus and the bronchial tree. Of all 13 patients, the postoperative period in terms of esophagotracheal complications, respiratory function, management, and outcome were analyzed. The median age of the patients was 57 years (range 29 to 72 years, mean 58.7 years). Eleven patients were male and two patients female. Six patients (46.1 %) presented with a squamous cell carcinoma and seven patients (53.9 %) with an adenocarcinoma (Table 1). Tumors were staged according to the TNM classification of the Union for International Cancer Control staging system [7]. Due to a locally advanced cancer, seven patients were treated with combined neoadjuvant radiochemotherapy and 1 patient had only neoadjuvant chemotherapy, according to a standardized protocol [8, 9]. After preoperative risk assessment, all patients underwent a transthoracic en-bloc esophagectomy with 2-field lymphadenectomy. Reconstruction was done with a gastric tube and an intrathoracic esophagogastrostomy (Ivor-Lewis esophagectomy). The surgical procedure is explained in detail elsewhere [10, 11]. Postoperative management Leakage of anastomosis was classified according to Low et al. into type I–III depending on the treatment of the leakage [2]. Type I leakage does not require any change in therapy or is treated medically, whereas type II leakage is defined by the necessity of interventional therapy (i.e., stent, EndoVac system, or drainage) and type III leakage Table 1 Characteristics of 13 patients with tracheo- or bronchoesophageal fistula after Ivor-Lewis esophagectomy No.
by the need of re-operation. Postoperative complications were graded I–V following the Dindo–Clavien Classification [12]. Pneumonia was defined using the Utrecht Scoring System for pneumonia following esophagectomy [13]. In case of a suspected leakage, an endoscopy for checkup of esophagogastric anastomosis was initiated as standard procedure. Endoscopic treatment consisted of an esophageal stent placement, either with a self-expending, covered metal stent (Ultraflex, Boston Scientific, Natick, MA, USA) or a self-expending, covered silicone stent (Polyflex, Boston Scientific). Bronchoscopy was conducted when TBF was strongly suspected (i.e., respiratory and/or cardiovascular failure, high leakage volume, preceding anastomotic leakage). Closure of TBF via bronchoscopy was realized by implantation of a self-expandable covered Y-stent (Leufen aerstent, Leufen Medical, Aachen, Germany). Re-thoracotomy was mandatory when a progressive septic course accompanied by multiple organ failure occurred or an ischemic gastric conduit was diagnosed. Depending on the intraoperative vascularization of the gastric conduit, the gastric tube was resected or re-anastomosed to the esophageal remnant. The tracheobronchial defect was covered with a pediculated pericardial or muscle flap (left sternocleidmastoideus muscle). The collected data were documented using Microsoft Excel 2010Ò. Statistical analysis was done by descriptive means only. PubMed database was searched for keywords ‘Ivor-Lewis esophagectomy,’ ‘tracheobronchial fistula,’ ‘anastomotic leakage,’ ‘gastrotracheal fistula,’ and ‘esophagobronchial fistula’ in order to compare the clinical data with previous reports.
Results
Patients and tumor characteristics Age
ASA
pTNM
Tumor entity
RTX/CTX
1
45
2
ypT0N0M0
AC
Yes
2 3
69 66
2 2
ypT1N0M0 ypT1N0M0
SCC SCC
Yes Yes
4
72
2
pT3N1M0
AC
No
5
70
3
pT1N0M0
SCC
No
6
57
2
ypT2N0M0
SCC
Yes
7
66
2
ypT3N1M0
SCC
Yes
8
68
2
pT3N0M0
AC
No
9
29
2
pT1N0M0
AC
No
10
55
2
ypT3N0M0
AC
Yes
11
54
3
pT1N0M0
AC
No
12
57
1
ypT1N0M0
AC
Yes
13
55
1
ypT3N1M0
SCC
Yes
AC adenocarcinoma, RTX/CTX radiochemotherapy, SCC squamous cell carcinoma
123
Postoperative complications In none of the 13 patients with TBF, an intraoperative injury of the tracheobronchial system was documented. All 13 patients were diagnosed to have an anastomotic leakage of the esophagogastrostomy either prior to the diagnosis of TBF (metachronous TBF, n = 6, 46.2 %) or simultaneously (synchronous TBF, n = 7, 53.8 %). Table 2 shows the clinical presentation in detail. A synchronous TBF was diagnosed by bronchoscopy in seven patients. TBF of the left main bronchus was the most frequent localization (6 patients, 46.1 %). Five patients (38.5 %) presented with a fistula into the trachea and two patients (15.4 %) into the right main bronchus. Synchronous TBF was located either in the left main bronchus (n = 4; 57.1 %) or the trachea (n = 3; 42.9 %). The size of the fistula varied from 3 to 15 mm in diameter (median
World J Surg Table 2 Clinical presentation and complications of 13 patients with tracheo- or bronchoesophageal fistula after Ivor-Lewis esophagectomy No
Esophagotracheal complications
Sepsis/MOF
Anastomotic Leakage (POD)
Type of leakage (I–III)a [2]
TBF (POD)
Localization of TBF
Size of TBF (mm)
Respiratory failure mechanical ventilation [72 h
Reintubation
Tracheostomy
Cumulative days of mechanical ventilation (days)
Cardiovascular failure
1
12
III
12
LMB
4
Yes
Yes
Yes
21
Yes
2
3
III
19
LMB
10
Yes
Yes
Yes
22
No
3
18
II
18
T
3
Yes
Yes
Yes
23
No
4
8
II
8
LMB
5
Yes
Yes
Yes
13
No
5
3
III
18
RMB
10
Yes
Yes
Yes
16
Yes
6 7
8 26
II II
37 27
LMB T
8 n/a
Yes Yes
Yes Yes
Yes Yes
94 7
Yes Yes
8
2
III
22
RMB
n/a
Yes
Yes
Yes
20
Yes
9
9
II
9
T
4
No
No
No
0,4
No
10
12
II
12
LMB
5
Yes
Yes
Yes
8
Yes
11
6/281
II
281
LMB
9
No
No
No
0,6
No
12
6
II
14
T
n/a
No
No
No
0,4
No
13
6
II
56
T
15
Yes
Yes
Yes
73
Yes
CRP C-reactive protein, LMB left main bronchus, n/a not available, PCT procalcitonin, POD postoperative day, RMB right main bronchus, T trachea, TBF tracheo- or bronchoesophageal fistula a
At the time of TBF diagnosis; at the time of first detection
6.5 mm). The majority of the patients (n = 10; 76.9 %) had to be re-intubated due to acute respiratory failure. In all ten patients, prolonged ventilation eventuated (cumulative mechanical ventilation: median 47 days, range 7–94) and a tracheostomy was performed. All patients developed a severe pneumonia in means of Utrecht pneumonia scoring C2 with pulmonary radiography findings, elevated leucozyte count (median 16.1 9 109/l, range: 8.4–33.2), and fever. With regard to the Dindo–Clavien classification, grade IIIa complications were recorded in four patients (30.8 %), grade III b and grade IVa in two patients, respectively (15.4 %). The hospital mortality (grade V complication) was 38.4 % (5 of 13 patients). Postoperative management Solitary endoscopic treatment was carried out in four patients (no. 3, 9, 10, and 12) and consisted of inserting an esophageal stent with the intention of simultaneous closure of both, anastomotic leakage and TBF (Table 3). This procedure was successful in 3 out of 4 patients (no. 3, 9, and 12). All three patients had fistulization to the trachea. Solitary endoscopic treatment of TBF was not successful in one patient (no. 10) with fistulization into the left main bronchus who suffered from a pneumonic sepsis and finally multiple organ failure. The patient died on postoperative day (POD) 20.
Four patients (no. 4, 7, 8, and 13) received endoscopic intervention with subsequent surgical re-exploration during the postoperative course. Two patients had endoscopic intervention with fibrin glue application (no. 4) or implantation of an esophageal EndoVac system (no. 7) because of anastomotic leakage with synchronous TBF. The EndoVac system describes the endoluminal placement of a vacuum-sealed sponge at the site of the esophageal anastomotic leakage. This sponge is connected to an external pump via a nasogastric tube applying a negative pressure. It has to be replaced every 3–4 days until complete healing of the anastomosis. Because of progressive respiratory failure in both cases, re-thoracotomy with resection of the gastric conduit and construction of a cervical esophagostomy was done. The bronchial defect was sutured and reinforced with a pediculated pericardial flap. Patient no. 8 underwent re-thoracotomy with re-anastomosis because of an anastomotic failure on POD 2. Due to a recurrent anastomotic leakage 18 days later the gastric conduit was finally resected and a cervical esophagostomy constructed. A concomitant laceration of the right main bronchus was closed with a primary suture. Two days later, a recurrent leakage of the right main bronchus was diagnosed and a pediculated pericardial flap was used to reinforce the bronchial suture line. The patient finally died on POD 157 following a cardiac arrest. Patient no. 13 received esophageal stent placement on POD 6 because of an
123
World J Surg Table 3 Synopsis of management and outcome of 13 patients with TBF No
Treatment of anastomotic leakage and TBF
Postoperative course
Outcome
Endoscopic yes/no (POD)
Bronchoscopic yes/no (POD)
Surgery yes/no (POD)
Procedures (POD)
ICU stay after esophagectomy (in days)
Discharge from hospital (POD)
Hospital mortality (POD)
Followup
1
No
Yes (12/13)
Yes (13)
Y-stent trachea (12); RT, RA, pPF, Y-stent trachea (13)
37
56
No
Alive
2
No
Yes (20)
Yes (3)
RT, RA (3); Y-stent trachea (20)
55
85
No
Alive
3
Yes (18)
No
No
Esophagus stent (18)
30
88
No
Alive
4
Yes (8)
No
Yes (29)
Fibrin glue (8); RT, GR, tEG, pPF (29)
55
106
No
Dead
5
No
No
RT, RA (4); RT, GR, tEG, pPF (19)
19
–
Yes (20)
Dead
6
Yes (8)
Yes (38)
No
Esophagus stent (8), Stent left main bronchus (38)
103
–
Yes (109)
Dead
7
Yes (26)
No
Yes (27)
Esophagus EndoVac (26), RT, GR, tEG, pPF (27)
57
90
No
Alive
8
Yes (20)
No
Yes (2/22/ 24)
RT,RA (2), Esophagus stent 20), RT, GR, suture of fistula (22), RT, pPF, fibrin glue (24)
131
–
Yes (130)
Dead
9
Yes (9)
No
No
Esophagus stent (9)
6
53
No
Alive
10
Yes (12)
No
No
Esophagus stent (12)
9
–
Yes (20)
Dead
11
Yes (6/281)
Yes (281)
No
Esophagus stent (6/281), Y-stent trachea (281)
20
61
No
Alive
12
Yes (6/14)
No
No
Esophagus stent (6), Replacement of Esophagus stent(14)
15
50
No
Alive
13
Yes (6)
No
Yes (56)
Esophagus stent (6), RT, RA, pSF (56)
74
–
Yes (79)
Dead
n = 10 (76.9 %)
n = 4 (30.8 %)
n=7 (53.8 %)
37 days (6–131)
73 days (50–106)
Yes n = 5 (38.5 %)
Yes (4/19)
GR resection of gastric conduit, ICU intensive care unit, POD postoperative day, pPF pediculated pericardial flap, pSF pediculated sternocleidmastoideus flap, RA re-anastomosis of esophagogastric anastomosis, RT re-thoracotomy, TBF tracheo- or bronchoesophageal fistula, tEG temporary cervical esophagostomy
anastomotic leakage. On POD 56, removal of esophageal stent revealed a TBF to the trachea. After left cervical incision with upper sternal split, the cervical defect of the trachea was primarily sutured and reinforced with the interposition of the mobilized left sternocleidmastoideus muscle. Despite this surgical intervention, the patient died on POD 79. Endoscopic esophageal stenting as well as bronchoscopic stenting (double stenting) was performed in two patients (no. 6 and 11). After initial double stenting (no. 6), a later endoscopic control of the esophageal stent revealed an enlargement of the anastomotic defect to half of the circumference as well as the TBF. The patient died without possible surgical options on POD 109. Patient no. 11 had a history of early anastomotic leakage on POD 6 which was successfully closed of an esophageal stent. After stent removal, multiple pneumatic dilatation of the stenotic
123
esophagogastrostomy had to be performed and re-insufficiency of anastomosis with a secondary TBF into the left main bronchus was visualized on POD 281. The patient received an interventional treatment with insertion of a second esophageal stent combined with a tracheal Y-stent. Two patients (no. 1 and 2) were treated by bronchoscopic intervention and surgery. In patient no. 1 who received a tracheal Y-stent for TBF repair on POD 12, a failure of mechanical ventilation due to a high-volume air leakage led to emergency surgery on the following day consisting of re-thoracotomy, re-anastomosis, and pediculated pericardial flap for fistula closure as well as reinsertion of a tracheal Y-stent. The re-operation was done using an extracorporal membranous oxygenation (ECMO). Patient no. 2 had re-thoracotomy with re-anastomosis of esophageal anastomosis because of early anastomotic leakage on POD 3. Eighteen days later, bronchoscopy
World J Surg
revealed a 10-mm-wide defect of the left main bronchus. A tracheal Y-stent was inserted, leading to a successful closure of TBF. The last patient (no. 5) of this series was treated only by surgical re-exploration. After development of early anastomotic leakage with following re-thoracotomy and reanastomosis on POD 4, bronchoscopy showed a fistula to the right main bronchus. Failure of mechanical ventilation made re-thoracotomy mandatory, and a gastrointestinal stent insertion was impossible due to ischemia of the gastric conduit. Resection of the ischemic gastric tube and a cervical esophagostomy was performed. The leakage of the right main bronchus was covered with a pediculated pericardial flap. The patient died 1 day after re-operation.
Discussion Transthoracic esophagectomy remains the best surgical option for cure of esophageal cancer and is therefore established as a standard treatment in most specialized centers for esophageal surgery [14]. Nevertheless, transthoracic esophagectomy with gastric reconstruction continues to be a challenging procedure since it is still accompanied by a considerably high rate of postoperative morbidity accounting to 40–50 % [1, 15–17]. The occurrence of TBF following transthoracic esophagectomy is rare, accounting to a reported incidence of 0.8–3.9 % [3, 5, 18, 19]. However, TBF is related with the highest mortality rate of all major complications reaching up to 57 % (Table 4). This retrospective study confirms these two basic features of TBF with an incidence of 1.1 % and a mortality rate of 38.5 %. At present, the etiology of TBF is not completely understood, but some confounding factors seem to trigger the development of TBF and support the idea of a multifactorial process. Dissection of smaller bronchial arteries may induce segmental ischemia of the tracheobronchial tree. This partial devascularization may occur after extensive peritracheal and peribronchial nodal dissection as part of the routinely performed mediastinal lymphadenectomy [3, 20]. Segmental ischemia of the vulnerable pars membranacea may be aggravated by the intraluminal pressure either by the cuff of the double-lumen tube in the left main bronchus during intraoperative single-lung ventilation or the tracheal cuff of a standard tube during prolonged mechanical ventilation postoperatively. In addition, a direct injury of the posterior aspect of the tracheobronchial tree mainly by electrocautery may contribute to a localized disturbance of microcirculation. The analysis of these retrospective data reveals another causal factor of TBF. As observed in other series [5, 18, 21], all our patients with TBF had an additional leakage of the
esophagogastrostomy. Since twelve patients had a failure of the circular and one patient of the longitudinal stapler line, it appears that not the anastomotic leakage itself but the accompanying mediastinitis with its inflammatory process next to tracheobronchial tree is the causal agent for the development of TBF [5]. The incidence of postoperative anastomotic failure is higher than that of TBF which means that not all patients with anastomotic leakage will necessarily develop a TBF. Therefore, the pathophysiological mechanism of TBF is most likely to be that of an intraoperatively induced ischemic damage of the pars membranacea with an additional postoperative inflammation of the mediastinum. In contrast to other series [3, 19], a direct injury of the trachea or the main bronchi with the need for a change of intraoperative management was not seen even in patients with SCC located at the level of tracheal bifurcation or above. This might be due to an exclusion of patients with a suspected T4 category who are at high risk of tracheobronchial injury during esophageal dissection. Therefore, it is of clinical importance that in all patients of this series the ischemic injury of the tracheobronchial system was not obvious during esophagectomy and the surgeon was generally not aware of the risk factor ‘ischemic damage.’ Several previous reports have tried to classify the localization of the TBF depending on the underlying cause. Yasuda et al. described ten patients with TBF following transthoracic esophagectomy and classified these cases into an ‘anastomotic leakage type’ developing in the trachea, the ‘gastric necrosis type’ observed predominantly around the carina, and the ‘gastric ulcer type’ associated with a fistula in the right main bronchus [6]. Bartels et al. classified TBF into an ‘ischemic’ and ‘non-ischemic’ type. The ‘ischemic’ type was localized around the carina and both main bronchi, and the ‘non-ischemic’ type was exclusively located in the trachea [3]. In this study, two predominant localizations of TBF could be identified and TBF could be classified with respect to the topographic relation to the anastomotic site into a direct (tracheal) and an indirect (left main bronchus) type. Only 2 out of 13 fistulas were located in the right main bronchus. The indirect fistulae to the left main bronchus predominantly present as metachronous TBF. Management of TBF is still difficult because this is a rare complication, and even for high-volume centers it is laborious to gain sufficient experience to modify its management and to implement a treatment algorithm. In addition, several completely different treatment strategies are under discussion all of which do not predict individual outcome at present. Therefore, treatment of TBF is still an individual decision making. However, based on the data of this retrospective study and previous reports some general recommendation for management of TBF can be made. The predominant clinical feature of TBF is acute
123
123
2015
Cologne
13
6
7
11 10
2
6
31
Patients with TBF after esophagectomy
18 (8–281)
25 (14–30)
n/a
n/a 13 (8–35)
n/a
30 (18–425)
12 (1–30)
Days to TBF median (range)
13 (100 %)
4 (66.6 %)
7 (100 %)
n/a 6 (60 %)
2 (100 %)
6 (100 %)
7 (22.5 %)
Anastomotic leakage
5 (38.5 %)
5 (83.3 %)
3 (42.8 %)
n/a 9 (90 %)
0 (0.0 %)
5 (83.3 %)
19 (61.3 %)
6 (46.1 %)
1 (16.7 %)
0 (0.0 %)
n/a 1 (10 %)
0 (0.0 %)
0 (0.0 %)
4 (12.9 %)
2 (15.4 %)
0 (0.0 %)
4 (57.2 %)
n/a 0 (0.0 %)
2 (100 %)
1 (16.7 %)
8 (25.8 %)
* Mortality rate with regard to the total number of 35 patients
0 (0.0 %)
2 (33.4 %)
1 (14.3 %)
0 (0.0 %) 4 (40 %)
0 (0.0 %)
2 (33.4 %)
n/a
Conservative treatment
RMB
Trachea
LMB
Treatment TBF
Localization of TBF
LMB left main bronchus, n/a not available, RMB right main bronchus, TBF tracheo- or bronchoesophageal fistula
2012
2010 2012
Shen et al. [4] Yasuda et al. [6]
2015
2002
Mangi et al. [21]
Morita et al. [19]
2001
Buskens et al. [18]
Schweigert et al. [5]
1998
Bartels et al. [3]
Date
Table 4 Synopsis of 86 cases of tracheobronchial fistula following Ivor-Lewis esophagectomy reported in the literature
4 (30.8 %)
0 (0.0 %)
2 (28.6 %)
0 (0.0 %) 0 (0.0 %)
0 (0.0 %)
0 (0.0 %)
n/a
Bronchoscopic treatment
7 (53.8 %)
4 (66.6 %)
2 (28.6 %)
11 (100 %) 6 (60 %)
2 (100 %)
4 (66.6 %)
n/a
Surgical treatment
5 (38.5 %)
0 (0.0 %)
4 (57.1 %)
2 (5.7 %*) 3 (30 %)
0 (0.0 %)
0 (0.0 %)
10 (32.2 %)
Hospital mortality
World J Surg
World J Surg
respiratory failure and the vast majority of patients with TBF require long-time mechanical ventilation with the need for tracheostomy. Progressive respiratory failure is primarily caused by two conditions, first the continuous aspiration through the fistula with subsequent pneumonia defined by the Utrecht score [13] and secondly an increasing air leakage which makes sufficient oxygenation difficult or even impossible. The third condition that has to be taken into consideration is the defect of the esophagogastric anastomosis and the vascularization of the gastric conduit. It is obvious that for patients with partial or complete necrosis of the gastric tube conservative (interventional) therapy is not a possible option. These patients definitely require re-thoracotomy with resection of the gastric conduit or re-do of the anastomosis. Although the re-anastomosis in the upper mediastinum or even upper thoracic inlet is a technically demanding procedure and depends on the length, the wall condition, and vascularization of the conduit, it should be the surgical therapy of choice since the gastric conduit aids to cover the tracheal or bronchial defect. In case of a conduit resection, intraoperative management of the functional air leakage is more difficult since the defect is only covered with a pediculated muscle or pericardial flap. This is particularly the case for a fistula of the left main bronchus in which the direct contact of the cuff of the double-lumen tube may enlarge the defect and suturing of any kind of flap closely to the defect is almost impossible. Patients with progressive respiratory failure due to an uncontrollable air leakage are the second patient group requiring re-thoracotomy. The principles of surgical management are similar to those patients having surgery for poor vascularization of the gastric conduit. In order to avoid this critical intraoperative condition, ventilation management includes the reduction of the airway pressure as much as possible aiming to achieve spontaneous ventilation via tracheostomy. For all other patients who do not require surgical therapy, the primary goal of treatment is the closure of the fistula in order to prevent progressive aspiration pneumonia with deterioration of respiratory function. Since the reflux from the gastrointestinal site is the etiologic agent of progressive pneumonia, stenting of the anastomotic leakage appears to be the interventional therapy of choice. This treatment option is well accepted for a solitary anastomotic leakage [22] but has also been suggested by other groups for treatment of TBF [5]. After insertion of a self-expandable stent covering the anastomotic defect, it is important to exclude persistent retrograde reflux by a CT scan with oral contrast medium applied via a nasogastric tube. As demonstrated in some patients of this series with a TBF of the trachea, treatment with a solitary gastroesophageal stent is feasible to achieve the primary goal of fistula closure. However, for these
patients with the tracheal type fistula double stenting is not indicated because the direct contact of the gastrointestinal and tracheal stent prevents granulation tissue to overgrow the existing defect. In this retrospective series, only four patients received some kind of a tracheal or bronchoscopic stent, always combined with either surgical or endoscopic treatment. Clinical management of these patients is rather difficult because the tracheobronchial stent acts as a foreign body and induces persistent pulmonary secretion. TBF is a rare but life-threatening complication with a high postoperative mortality rate. Fistulization predominantly goes to the left main bronchus or trachea and is triggered by intrathoracic leakage of the esophagogastrostomy and its inflammatory process of the mediastinum. Therefore, the treatment options primarily depend on the vascularization of the gastric conduit and the respiratory function defined by the severity of the aspiration pneumonia and the volume of the concomitant air leakage. Patients with anastomotic leakage and gastric necrosis always require immediate re-thoracotomy as well as patients with progressive respiratory failure. In all other patients, the closure of the fistula can be primarily tried to achieve from the gastrointestinal site with a self-expanding stent.
References 1. Sauvanet A, Mariette C, Thomas P et al (2005) Mortality and morbidity after resection for adenocarcinoma of the gastroesophageal junction: predictive factors. J Am Coll Surg 201:253–262 2. Low DE, Alderson D, Cecconello I et al (2015) International consensus on standardization of data collection for complications associated with esophagectomy: esophagectomy complications consensus group (ECCG). Ann Surg 262:286–294 3. Bartels HE, Stein HJ, Siewert JR (1998) Tracheobronchial lesions following oesophagectomy: prevalence, predisposing factors and outcome. Br J Surg 85:403–406 4. Shen KR, Allen MS, Cassivi SD et al (2010) Surgical management of acquired nonmalignant tracheoesophageal and bronchoesophageal fistulae. Ann Thorac Surg 90:914–918 discussion 919 5. Schweigert M, Dubecz A, Beron M et al (2012) Management of anastomotic leakage-induced tracheobronchial fistula following oesophagectomy: the role of endoscopic stent insertion. Eur J Cardiothorac Surg 41:e74–80 6. Yasuda T, Sugimura K, Yamasaki M et al (2012) Ten cases of gastro-tracheobronchial fistula: a serious complication after esophagectomy and reconstruction using posterior mediastinal gastric tube. Dis Esophagus 25:687–693 7. Wittekind C, Meyer HJ (2013) TNM: Klassifikation Maligner Tumoren, 7. Auflage 8. Cunningham D, Allum WH, Stenning SP et al (2006) Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med 355:11–20 9. van Hagen P, Hulshof MC, van Lanschot JJ et al (2012) Preoperative chemoradiotherapy for esophageal or junctional cancer. N Engl J Med 366:2074–2084
123
World J Surg 10. Holscher AH, Schneider PM, Gutschow C et al (2007) Laparoscopic ischemic conditioning of the stomach for esophageal replacement. Ann Surg 245:241–246 11. Schroder W, Holscher AH, Bludau M et al (2010) Ivor-Lewis esophagectomy with and without laparoscopic conditioning of the gastric conduit. World J Surg 34:738–743. doi:10.1007/s00268010-0403-x 12. Dindo D, Demartines N, Clavien PA (2004) Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 240:205–213 13. van der Sluis PC, Verhage RJ, van der Horst S et al (2014) A new clinical scoring system to define pneumonia following esophagectomy for cancer. Dig Surg 31:108–116 14. Rice TW, Rusch VW, Apperson-Hansen C et al (2009) World wide esophageal cancer collaboration. Dis Esophagus 22:1–8 15. McCulloch P, Ward J, Tekkis PP (2003) Mortality and morbidity in gastro-oesophageal cancer surgery: initial results of ASCOT multicentre prospective cohort study. BMJ 327:1192–1197 16. Bailey SH, Bull DA, Harpole DH et al (2003) Outcomes after esophagectomy: a ten-year prospective cohort. Ann Thorac Surg 75:217–222 discussion 222
123
17. Kutup A, Nentwich MF, Bollschweiler E et al (2014) What should be the gold standard for the surgical component in the treatment of locally advanced esophageal cancer: transthoracic versus transhiatal esophagectomy. Ann Surg 260:1016–1022 18. Buskens CJ, Hulscher JB, Fockens P et al (2001) Benign tracheoneo-esophageal fistulas after subtotal esophagectomy. Ann Thorac Surg 72:221–224 19. Morita M, Saeki H, Okamoto T et al (2015) Tracheobronchial fistula during the perioperative period of esophagectomy for esophageal cancer. World J Surg 39:1119–1126. doi:10.1007/ s00268-015-2945-4 20. Fujita H, Kawahara H, Hidaka M et al (1988) An experimental study on viability of the devascularized trachea. Jpn J Surg 18:77–83 21. Mangi AA, Gaissert HA, Wright CD et al (2002) Benign broncho-esophageal fistula in the adult. Ann Thorac Surg 73:911–915 22. Leers JM, Vivaldi C, Schafer H et al (2009) Endoscopic therapy for esophageal perforation or anastomotic leak with a self-expandable metallic stent. Surg Endosc 23:2258–2262
