Surgical Management of Acquired Nonmalignant Tracheoesophageal and Bronchoesophageal Fistulae

Transcription

1 Surgical Management of Acquired Nonmalignant Tracheoesophageal and Bronchoesophageal Fistulae K. Robert Shen, MD, Mark S. Allen, MD, Stephen D. Cassivi, MD, MS, Francis C. Nichols III, MD, Dennis A. Wigle, MD, PhD, W. Scott Harmsen, MS, and Claude Deschamps, MD Divisions of General Thoracic Surgery and Biostatistics, Mayo Clinic, Rochester, Minnesota Background. Acquired nonmalignant fistulae between the airway and esophagus (tracheoesophageal fistulae [TEF]) are rare life-threatening conditions. Several management approaches have been proposed, while the optimal strategy remains controversial. Methods. This study is a retrospective review of all patients with TEF treated at our institution from 1978 through Results. Thirty-five patients (22 men, 13 women) underwent surgical repair of acquired nonmalignant TEF. Median age was 55 years (range, 5 to 78). Most common causes were the following: complications of esophageal surgery (11), trauma (6), granulomatous infection (5), stent erosion (4), and prolonged mechanical ventilation (2). Location was proximal trachea in 7, mid-trachea in 5, and distal trachea or bronchus in 23. Fifty-six operations were performed. Six patients had staged repair, with 1 patient requiring 4 operations for recurrent TEF. TEF division and primary repair was performed in 18 patients, esophageal resection with reconstruction in 4, and esophageal diversion in 6. Four patients had suture closure of the esophageal or tracheal defect only, and 3 required segmental tracheal or bronchial resection. Four patients were ventilator dependent at the time of repair. Pedicled tissue flaps were used in 28 patients (80%). Operative mortality was 5.7% (2 of 35). Nineteen patients (54.3%) had complications. Median hospital stay was 14 days (range, 4 to 209). Median follow-up was 30.4 months (range, 0.5 to 233) and complete in 34 (97.1%). Three patients (8.6%) developed recurrent TEF. Twenty-nine patients resumed oral intake. One patient required a permanent tracheal T tube. Conclusions. Single-stage primary repair of both airway and esophageal defects with tissue flap interposition can safely be performed successfully in the majority of patients with acquired nonmalignant TEF. (Ann Thorac Surg 2010;90:914 9) 2010 by The Society of Thoracic Surgeons Acquired nonmalignant fistulization between the airway and esophagus is a rare but challenging clinical problem. The etiology of these tracheoesophageal fistulae (TEF) include complications of mechanical ventilation [1] or indwelling tracheal or esophageal stents [2], complications from prior tracheal or esophageal surgery [3 5], granulomatous mediastinal infections [6, 7], trauma [8, 9], iatrogenic injuries [10], and caustic ingestion. Regardless of the etiology, TEFs are a life-threatening condition due to the devastating pulmonary complications of ongoing tracheobronchial contamination and interference with nutrition. Several differing approaches to the management of TEFs have been proposed [11, 12] and the optimal management strategy is controversial. Direct suture closure of both the tracheal and esophageal defects [5, 13 16], segmental tracheal resection and primary anastomosis with direct esophageal closure [13, 15 18], tracheal closure using an esophageal patch [13], closure of the Accepted for publication May 17, Presented at the Forty-sixth Annual Meeting of The Society of Thoracic Surgeons, Fort Lauderdale, FL, Jan 25 27, Address correspondence to Dr Shen, Division of General Thoracic Surgery, Mayo Clinic, 200 First St, SW, Rochester, MN 55905; shen.krobert@mayo.edu. defects with soft tissue flaps [13, 14, 19], a combined surgical and endoscopic approach [20], and a two-stage approach with esophageal diversion and primary closure of the tracheal defect [10, 13, 15,] have all been advocated. The use of self-expanding esophageal metal stents to successfully manage TEF has also been reported in a small number of patients [21]. The aim of this study was to analyze our experience with surgical management of TEFs over a 30-year period and analyze our short-term and long-term outcomes. Patients and Methods Patient Selection All patients who were diagnosed with a TEF and treated at Mayo Clinic in Rochester, Minnesota between January 1, 1978 and December 31, 2007 were identified from a prospectively maintained surgical database. There were 246 patients treated for a TEF during this time frame. One hundred twenty-six patients (51.2%) had malignant TEFs from primary aerodigestive tract malignancies, or metastatic tumors to the airway or esophagus. Eighty-five patients (34.6%) were surgically treated for congenital TEFs, usually within the first few weeks of life. Thirty-five patients (14.2%) underwent surgical repair of an acquired 2010 by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc doi: /j.athoracsur

2 Ann Thorac Surg SHEN ET AL 2010;90:914 9 TREATMENT OF TRACHEOESOPHAGEAL FISTULAE 915 nonmalignant TEF and these 35 patients are the focus of this study. Characteristics of these 35 patients are shown in Table 1. Figure 1 shows the location and type of fistulae into the airway. The medical records were reviewed for demographic information, presenting symptoms, prior surgery, operative procedures, pathology, morbidity and mortality, length of hospitalization, and last follow-up visit or date of death. Operative mortality included all deaths occurring within 30 days of the operative procedure and those who died later but during the same hospitalization. Survival data not available in the medical record were obtained from the Social Security Death Index. Follow-up by chart review and with the primary care physician was performed. The date of the initial surgical repair of the TEF was the starting point in the survival estimation and the date of death or last follow-up the endpoint. The Mayo Foundation Institutional Review Board approved this study and the need for individual patient consent was waived. Table 1. Demographics and Characteristics of 35 Patients who Underwent Surgical Repair of a Nonmalignant Tracheoesophageal or Bronchoesophageal Fistula Characteristic Median Range Patient age 55 years 5 78 years Interval between initial event 6.3 months months that led to development of TEF and surgical repair Size of fistula 5.0 mm mm (No.) (%) Gender Male Female Etiology of fistula Complication of previous esophageal surgery Laryngotracheal trauma Granulomatous mediastinal infection Erosion of indwelling airway or esophageal stent Complication of prolonged mechanical ventilation Complication of radiation therapy Recurrence of congenital TEF 19 and years after initial surgical repair performed during first week of life Caustic lye ingestion Complication of endoscopic antireflux procedure Complication of Crohn s disease TEF tracheoesophageal fistula. Fig 1. Location of the airway fistula. Statistical Analysis Descriptive statistics for categoric variables are reported as frequency and percentage while continuous variables are reported as mean (standard deviation) or median (range) as appropriate. Univariate logistic regression was used to assess the association between variables of clinical interest and postoperative complications after surgical repair of TEF. Factors assessed included age, gender, etiology of TEF, time period TEF was performed, location of the TEF, history of prior treatment of TEF, ventilator dependence at the time of TEF repair, whether repair was done as a multistaged procedure, surgical approach, type of surgical repair, and use of a pedicled tissue flap to reinforce the repair. A total of 19 patients had a postoperative complication, therefore multivariable analyses were not done. The SAS version 9.1 (SAS Institute Inc, Cary, NC) was used for all analyses. Results Prior Treatment Eight patients (22.8%) had undergone prior TEF treatment before they underwent surgical treatment at our institution. The two patients with recurrent TEF after prior repair of congenital TEF were both initially treated with division of the TEF, suture closure of the tracheal defect, and primary esophageal anastomosis. Two patients had undergone esophageal diversion with cervical esophagostomy. Two patients had undergone failed attempted surgical repair with recurrent TEF. One patient had undergone attempt to repair the TEF with endoscopic closure and one patient had undergone placement of an esophageal stent. Surgical Treatment Fifty-six operations were performed in 35 patients (Fig 2). Twenty-eight patients underwent a single operation to repair the fistula. Seven patients underwent multistaged repair of their TEF. Four patients underwent a two-stage repair and 2 underwent three-staged operations to repair their TEF. One patient with a TEF resulting from complications of Crohn s disease was operated on 4 times for recurrent TEF. Ten operations were performed on 8 patients to treat complications after initial surgical repair of the TEF. Four patients required tracheostomy and 4 were reexplored for cervical bleeding-hematoma. One patient required esophageal diversion 9 days after failure

3 916 SHEN ET AL Ann Thorac Surg TREATMENT OF TRACHEOESOPHAGEAL FISTULAE 2010;90:914 9 Table 3. Complications of Surgery a Complication No. % Fig 2. Breakdown of all of the operative procedures performed. (pt patients.) of the initial primary repair of the tracheal and esophageal defects. Gastrointestinal continuity was restored 11 months later using a substernal stomach conduit. One patient underwent talc pleurodesis to treat persistent chylothorax after TEF repair. The surgical approach used to repair the fistulae included a right posterolateral thoracotomy in 20 patients (57.1%), low collar cervical incision in 9 (25.7%), a combined cervical and abdominal approach in 2 (5.7%), a right chest and abdominal approach in 2 (5.7%), and a left thoracotomy and a thoracoabdominal approach in 1 patient each (2.9%). A singlestage direct suture closure of both the tracheal and esophageal defect was used in 18 patients (51.4%). Esophageal diversion was performed in 6 patients (17.1%), and in 4 (11.4%) gastrointestinal continuity was never reestablished. One patient had a staged colon interposition and one patient had a composite colon and jejunal graft to reestablish gastrointestinal continuity as a second staged procedure. Four patients (11.4%) underwent suture closure of the tracheal defect with esophageal resection and immediate reconstruction. Two patients (5.7%) had suture closure of the esophageal defect, with muscle flap interposition to patch the tracheal defect. Two patients (5.7%) underwent suture closure of the tracheal defect with esophageal resection and delayed esophageal reconstruction as a second stage operation. Segmental tracheal or bronchial resection and reconstruction was performed in 3 patients (8.6%). Four patients were ventilator dependent at the time of the initial surgical repair of the TEF. Interposed pedicled tissue flaps were used in Table 2. Soft Tissue Interposition Flaps Used to Separate Esophagus and Trachea Flap No. % Serratus anterior muscle Sternocleidomastoid muscle Latissimus dorsi muscle Intercostal muscle Pleural flap Pericardial fat Omohyoid muscle Pectoralis major muscle Sternohyoid muscle Azygous vein Total Respiratory failure Pneumonia Esophageal leak Postoperative bleeding requiring reoperation Recurrent TEF Tracheal dehiscence Bacteremia Atrial fibrillation Vocal cord paralysis Prolonged airleak Wound dehiscence ARDS Clostridium difficile Empyema Pancreatitis Heparin-induced thrombocytopenia Wound infection Chylothorax a Some patients experienced greater than 1 complication. TEF tracheoesophageal fistulae. 28 patients (80%). The types of flaps that were used are shown in Table 2. Morbidity and Mortality Nineteen patients (54.3%) had postoperative complications and 8 patients (22.8%) required reoperation for complications. The list of postoperative complications is shown in Table 3. Respiratory failure and pneumonia were the most common postoperative complications. Univariate logistic regression was used to assess the association between variables of clinical interest and increased risk of postoperative complications. Factors analyzed included age, gender, etiology of fistula, time from initial event to development of TEF, surgical era in which repair was performed, location of fistula, fistula size, prior treatments, single versus multistaged repair, surgical approach, type of repair, use of soft tissue flaps, and ventilator dependence. None of the assessed factors were associated with increased risk of postoperative complications. A separate univariate logistic regression analysis of whether the size of the fistula was associated with increased need for reoperation, need for esophageal diversion, or increased length of hospitalization showed no significant association. Univariate logistic regression was also used to assess the association between variables of clinical interest and whether direct repair of the fistula was performed or esophageal diversion was required. Factors analyzed included age, gender, etiology of fistula, time from initial event to development of TEF, surgical era in which repair was performed, location of fistula, fistula size, prior treatments, single versus multistaged repair, surgical approach, type of repair, use of soft tissue flaps, and ventilator dependence. None of the assessed factors were

4 Ann Thorac Surg SHEN ET AL 2010;90:914 9 TREATMENT OF TRACHEOESOPHAGEAL FISTULAE 917 associated with increased need to perform esophageal diversion rather than direct repair of the fistula. Two patients (5.7%) died after repair of TEF. Both patients had early failure of the surgical repair with recurrent TEF. One died of respiratory failure on postoperative day 30 and the other patient died on postoperative day 209 of multisystem organ failure as a result of ongoing airway contamination. Three patients developed recurrent TEF (8.6%). One patient with a history of Crohn s disease underwent 4 unsuccessful operations for recurrent TEF. The two patients with recurrent TEF never left the hospital and died of complications related to recurrent TEF. The median length of hospitalization was 14 days (range, 4 to 209 days). Median follow-up was 30.4 months (range, 0.5 to 233 months) and was complete in 34 patients (97.1%). Twenty-nine of the patients (82.9%) were able to return to an oral diet. Eight patients developed strictures of the airway or esophagus. The median time to development of the stricture was 319 days (range, 11 to 778 days). The median number of dilations required was 2 (range, 1 to 15 dilations). Six patients (17.1%) developed esophageal strictures only, 1 (2.9%) developed both tracheal and esophageal strictures, and 1 (2.9%) developed a tracheal stricture only. The patient with the isolated tracheal stricture developed TEF as a complication of an infected pseudoaneurysm of the innominate artery after mediastinoscopy. At the time of the TEF repair, the esophagus was closed primarily and the posterior membranous tracheal defect was covered with a pectoralis major flap. The patient subsequently developed tracheal stricture and tracheomalacia, which was successfully managed with a tracheal silastic stent. Ultimately the stent was able to be removed after 7 months. One patient required a permanent tracheal appliance to provide an airway. This patient had developed TEF as a complication of a percutaneous tracheostomy tube placement and had undergone primary suture repair of the tracheal and esophageal defects at an outside hospital that failed. He underwent primary closure of the esophageal defect and segmental tracheal resection and reconstruction at the Mayo Clinic 21 months later. On postoperative day 7 he developed tracheal dehiscence and underwent reoperation. Cervical skin flaps were used to close the posterior membranous tracheal defect and a permanent tracheostomy tube was placed through the anterior tracheal defect. Comment Over a period of 30 years, 35 patients have undergone surgical repair of acquired nonmalignant fistula between the airway and esophagus at the Mayo Clinic Rochester. In contrast to the other single institution series of acquired nonmalignant fistulae, the most common etiologies for the fistulae in our series were complications of prior esophageal surgery and trauma. These differences in the etiology of the fistulae are also reflected in the location of the fistulae and the surgical approach used in our series. Nine patients in our series had distal tracheal fistulae and 14 patients had bronchoesophageal fistulae. The most common surgical approach used to repair the fistulae was a right thoracotomy, and the most common operative repair was direct primary suture closure of both the tracheal and esophageal defect with pedicled soft tissue flap interposition. Operative mortality in our series was 5.7%, which is similar to what other experienced centers have found. Operative mortality rates of 3.2% to 29.6% have been reported [13 16, 18]. Three patients (8.6%) in our series developed recurrent TEF. Recurrence rates of 6.8% to 7.9% have been reported by other centers with a comparable number of patients [14 16]. Twenty-nine of the 33 surviving patients in our series were able to return to an oral diet, and only 1 required a permanent airway appliance. There is a diversity of opinion regarding the optimal surgical management of TEFs. Whether the fistula can be simply divided and the tracheal and esophageal defects closed primarily or whether segmental tracheal or bronchial resection and reconstruction should be performed is controversial. Published results of the repair of acquired nonmalignant TEF are limited to the experience of only a few institutions. Mathisen and colleagues [16] and Hilgenberg and Grillo [17] described a single-stage technique for simultaneous two-layer closure of the esophagus, circumferential tracheal resection, and primary tracheal anastomosis once patients have been weaned from mechanical ventilatory support. Mathisen and colleagues have reported their results with this approach on 38 patients and Hilgenberg and Grillo reported results on 20 patients. In the series by Mathisen and colleagues of 41 operations on 38 patients, nine patients (23.7%) were managed by simple division and closure of the fistula, whereas tracheal resection and reconstruction was performed in 29 patients (76.3%). A cervical incision was used in 26 patients and a lateral thoracotomy was only used in 4 patients. Ten patients had a partial sternotomy in addition to a cervical incision to provide distal exposure, and 1 patient had a full median sternotomy in addition to a cervical collar incision. The mortality rate was 10.9%, which was mostly related to the ongoing need for mechanical ventilation. There were three recurrent fistulae and one patient developed delayed tracheal stenosis, all of which were managed successfully. Of the 34 survivors, 33 patients alimented orally and 32 were able to breathe without a tracheal appliance. In the 35 patients who were treated surgically for acquired nonmalignant TEF at our institution, only 3 patients (8.6%) underwent segmental tracheal or bronchial resection and 18 (51.4%) were successfully treated with a single-stage division of the fistula and direct suture repair of both the tracheal and esophageal defect. One possible explanation for this difference in approach is that in the patients reported on by Mathisen and colleagues [16], the vast majority (27 of 31 patients or 87.1%) of the TEFs resulted from the combined effects of pressure from either an endotracheal tube or a tracheostomy tube and a nasogastric tube, with resulting circumferential pressure necrosis of the tracheal wall. The presence of circumferential damage to the trachea and

5 918 SHEN ET AL Ann Thorac Surg TREATMENT OF TRACHEOESOPHAGEAL FISTULAE 2010;90:914 9 increased size of the fistula make segmental tracheal resection and reconstruction more likely to be required in cases where the TEF results as a complication of cuffed tracheal tubes. Macchiarini and colleagues [18] have also concluded, after having tried both techniques, that postintubation TEFs are best treated with segmental tracheal resection and reconstruction and primary esophageal closure rather than division and direct closure of both defects. In our series, the etiology of the TEF was more varied. Complications of esophageal surgery, laryngotracheal trauma, granulomatous infection, and stent erosion were the most common causes of TEF. Only two of the patients treated in our series developed TEF as a result of prolonged ventilation. Furthermore, the TEFs that were treated in our patients also had a higher distribution of distal tracheal or bronchial involvement (23 of 35 patients). Several other series have also reported that division of the fistula and direct suture closure of the tracheal and esophageal defects were the most common successfully employed techniques used to repair acquired nonmalignant TEFs. Baisi and colleagues reported on 31 patients who were treated for acquired benign TEFs [14]. Division of the fistula and closure of the tracheal and esophageal defect was performed in 26 patients and one patient underwent tracheal resection and reconstruction and primary esophageal closure. Operative mortality was 3.2%, and recurrence of the TEF occurred in 2 patients (6.4%). Twenty-three of the 31 patients (74.2%) had excellent long-term results and with normal function of both the esophagus and airway. Cherveniakov and colleagues [15] treated 31 patients surgically with acquired benign TEFs over a 25-year period. Simple excision of the fistula and suture closure of the trachea and esophagus was the method of choice in 21 patients. Six patients were treated with segmental tracheal resection and reconstruction with suture closure of the esophagus and four required esophagectomy with colon graft reconstruction. Operative mortality was 6.4% and no patients developed recurrent TEF. Twenty-nine of the 31 patients (93.5%) had good or excellent long-term results. Another controversy in managing acquired nonmalignant TEFs concerns the timing of fistula closure in ventilator-dependent patients. Hilgenberg and Grillo and Mathisen and colleagues have advocated an aggressive approach to control the effects of the fistula while stabilizing the patient, and weaning the patient from the ventilator before definitive one-stage surgical correction. Bartlett [12] and Thomas [11] have advocated attempting closure of the TEF while the patient is still receiving positive-pressure ventilation, with varying results. In the 35 patients we have treated, 4 patients were still ventilator-dependent at the time of surgical repair. Although we also favor delaying the single-stage repair of the TEF until the patient is no longer dependent on mechanical ventilation if possible, we do not consider ventilator dependence to be a contraindication to successful surgical repair of TEF, particularly in the case of patients in whom a segmental tracheal resection and reconstruction is not required. In conclusion, surgical management of patients with acquired nonmalignant fistulae between the airway and esophagus is challenging. Single-stage direct suture repair of both the tracheal and esophageal defect with interposed soft tissue flaps can be performed successfully in the majority of patients with acquired nonmalignant TEF. References 1. Flege JB Jr. Tracheoesophageal fistula caused by cuffed tracheostomy tube. Ann Surg 1967;166: Schowengerdt CG. Tracheoesophageal fistula caused by a self-expanding esophageal stent. Ann Thorac Surg 1999;67: Bartels HE, Stein HJ, Siewert JR. Tracheobronchial lesions following oesophagectomy: prevalence, predisposing factors and outcome. Br J Surg 1998;85: Iannettoni MD, Whyte RI, Orringer MB. Catastrophic complications of the cervical esophagogastric anastomosis. J Thorac Cardiovasc Surg 1995;110: Mangi AA, Gaissert HA, Wright CD, et al. Benign bronchoesophageal fistula in the adult. Ann Thorac Surg 2002;73: Gerzic Z, Rakic S, Randjelovic T. Acquired benign esophagorespiratory fistula: report of 16 consecutive cases. Ann Thorac Surg 1990;50: Macchiarini P, Delamare N, Beuzeboc P, et al. Tracheoesophageal fistula caused by mycobacterial tuberculosis adenopathy. Ann Thorac Surg 1993;55: Antkowiak JG, Cohen ML, Kyllonen AS. Tracheoesophageal fistula following blunt trauma. Arch Surg 1974;109: Kelly JP, Webb WR, Moulder PV, Moustouakas NM, Lirtzman M. Management of airway trauma. II: combine injuries of the trachea and esophagus. Ann Thorac Surg 1987;43: Gudovsky LM, Koroleva NS, Biryukov YB, Chernousov AF, Perelman MI. Tracheoesophageal fistula. Ann Thorac Surg 1993;55: Thomas AN. The diagnosis and treatment of tracheoesophageal fistula caused by cuffed tracheal tubes. J Thorac Cardiovasc Surg 1973;65: Bartlett RH. A procedure for management of acquired tracheoesophageal fistula in ventilator patients. J Thorac Cardiovasc Surg 1976;71: Marzelle J, Dartevelle P, Khalife J, Rojas-Miranda A, Chapelier A, Levasseur P. Surgical management of acquired postintubation trachea-oesophageal fistulas: 27 patients. Eur J Cardiothorac Surg 1989;3: Baisi A, Bonavina L, Narne S, Peracchia A. Benign tracheoesophageal fistula: results of surgical therapy. Dis Esophagus 1999;12: Cherveniakov A, Tzekov C, Grigorov GE, Cherveniakov P. Acquired benign esophago-airway fistulas. Eur J Cardiothorac Surg 1996;10: Mathisen DJ, Grillo HC, Wain JC, Hilgenberg AD. Management of acquired nonmalignant tracheoesophageal fistula. Ann Thorac Surg 1991;52: Hilgenberg AD, Grillo HC. Acquired nonmalignant tracheoesophageal fistula. J Thorac Cardiovasc Surg 1983;85: Macchiarini P, Verhoye JP, Chapelier A, Fadel E, Dartavelle P. Evaluation and outcome of different surgical techniques for postintubation tracheoesophageal fistulas. J Thorac Cardiovasc Surg 2000;119: Oliaro A, Rena O, Papalia P, et al. Surgical management of acquired non-malignant tracheo-esophageal fistulas. J Cardiovasc Surg (Torino) 2001;42: Freire JP, Feijó SM, Miranda L, Santos F, Castelo HB. Tracheo-esophageal fistula: combined surgical and endoscopic approach. Dis Esophagus 2006;19: Blackmon SH, Santora R, Schwarz P, Barroso A, Dunkin BJ. Utility of removable covered self-expanding metal stents for leak and fistula management. Ann Thorac Surg 2010;89:931 7.

6 Ann Thorac Surg SHEN ET AL 2010;90:914 9 TREATMENT OF TRACHEOESOPHAGEAL FISTULAE 919 DISCUSSION DR RAFAEL S. ANDRADE (Minneapolis, MN): I congratulate you for putting this together. I see that this is a very uncommon problem, which explains probably what most of us go through. We have about one patient a year like this and have learned a little bit from our experience. What we have learned is that when we get these patients they are pretty sick and they often need to be fine-tuned in order to get to a major operation; we have found the use of esophageal stents, never airway stents, useful in getting these patients temporized, in addition to nutrition and all that. Can you elaborate on your experience from that point of view? DR SHEN: In the last 10 years there has been increasing use of covered stents to address patients with a fistula between the airway and the esophagus. I really think their role is primarily for patients with malignant tracheoesophageal fistulas. In fact, it would probably be the treatment of choice for a patient with a malignant fistula. I believe their role for patients with acquired benign fistulas should be very limited. Placing a stent and hoping that this is going to resolve the problem without surgical intervention is wishful thinking. The scenario in which a stent may be useful as a temporizing measure is the patient with a benign fistula who is ventilator dependent. Patients with postintubation tracheoesophageal fistulas are more likely to have large circumferential defects and consequently much more likely to require a tracheal resection and reconstruction rather than direct closure. In these patients it is desirable to try to wean them off mechanical ventilation before undertaking a tracheal resection. In this scenario temporary placement of a stent to control ongoing tracheobronchial contamination is appropriate, but otherwise I would not favor the primary use of a covered stent on either the airway or the esophageal side for a benign fistula. DR DAVID TOM COOKE (Sacramento, CA): Rob that was an excellent review. For the patients presented to your group that were not in discontinuity, what were the patient characteristics and the treatment algorithm that led you to choose esophageal resection and reconstruction for the repair? DR SHEN: You re talking about the ones that we diverted or that we resected and reconstructed? DR COOKE: Resected and reconstructed. DR SHEN: Those were mostly patients with distal fistulas to either the bronchus or distal trachea, and many of those were patients where the etiology of the fistula itself was a complication of an esophagectomy. In many of these patients there was an anastomotic leak and, as one would imagine, there was a lot more inflammation and breakdown of the conduit itself that would not have lent itself to a primary repair. DR COOKE: So they weren t bronchoesophageal fistulas; they were bronchoenteric fistulas? DR SHEN: Yes many of them were fistulas from the bronchus or distal trachea to a gastric conduit after esophagogastrectomy. DR COOKE: I see. Thank you. DR WAYNE HOFSTETTER (Houston, TX): Dr Shen, very nice job on the review of the demographics and the outcomes of this group of patients. I want to take something home clinically from this. I m going to reiterate the question that was just asked. I think that the experience factor that comes into the operating room with the thoracic surgeons is whether to repair with a flap or whether to divert. So rather than doing a univariate and multivariate analysis looking at the overall survival, did you give consideration to a multivariate analysis of the decision-making process to go to diversion? I think that is one of the more interesting clinical questions that comes up. DR SHEN: The numbers are too small to do any sort of multivariate analysis either on the whole cohort or just the patients who were diverted. The number of patients that were diverted was low. Out of the total of 35 patients, only 6 were diverted. Of the six that were diverted, four patients never had gastrointestinal continuity reestablished. Our feeling is that most patients do not require an esophageal diversion and that should be avoided in order to eliminate the complexity of having to go back and reconstruct the gastrointestinal tract. Our experience has been that the majority of patients can be successfully treated with direct primary repair of both the tracheal and the esophageal defects. The other large series of these patients was 38 patients reported by the group from Massachusetts General Hospital [MGH] where, over a 20-year period of time, 38 patients underwent 41 operations. Seventy percent of those patients had tracheal resection and reconstruction, segmental airway resection, and reconstruction. I think one of the differences in our experience versus the MGH experience is that in most of those patients the etiology of the fistula was postintubation. In that group of patients there are problems with a larger, more circumferential airway injury. I think even in that group of patients, usually esophageal diversion was only used when there were no other options. DR THOMAS A. D AMICO (Durham, NC): So, Dr Shen, would you please address Dr Hofstetter s question. If you saw a patient with this, would you primarily divert them? DR SHEN: I would only divert them if I really thought that none of the other options were doable. It would be the option of last resort. DR RICHARD FREEMAN (Indianapolis, IN): I would just echo Dr Andrade s comment, not as an ultimate solution, but I think stenting in several people s practices has been found to be a good temporizing measure. My specific question is have you looked at consideration of placing an esophageal stent in the people who fail the primary operative repair? We have at least had some good experience in that. DR SHEN: Out of the 35 patients in our series, there was one patient, one of the patients who died, in whom that strategy was utilized; where an esophageal stent was placed following failed primary surgical repair to try to control the ongoing airway contamination. Ultimately the patient did succumb to the recurrence of the fistula and the complications. So to reiterate, I think stents in the nonmalignant situation may be helpful as a temporizing measure, but I don t think that it is going to be a definitive solution for the vast majority of these patients.