Tracheobronchial Injury in the Setting of an Esophagectomy for Cancer:

Postoperative Discovery a Bad Omen

VADIM P. KOSHENKOV, MD,1* DANNY YAKOUB, MD, PhD,2,3

ALAN S. LIVINGSTONE, MD,2,3 AND DIDO FRANCESCHI, MD2,3

1Division of Surgical Oncology, Rutgers Cancer Institute of New Jersey, New Brunswick, New Jersey2DeWitt Daughtry Family Department of Surgery, Division of Surgical Oncology, Miller School of Medicine, University of Miami, Miami,

Florida3Sylvester Comprehensive Cancer Center, Miller School of Medicine, University of Miami, Miami, Florida

Background: A tracheobronchial injury is an uncommon complication of an esophagectomy. Differences in outcomes may exist for patients withinjuries detected intraoperatively and postoperatively.Methods:A retrospective reviewwas performed for patients who underwent an esophagectomy for cancer at JacksonMemorial Hospital/Universityof Miami from January 2000 to June 2012.Results: An injury to the tracheobronchial tree occurred in 7 of 425 patients (1.6%). The majority of the operations were performed via a transhiatalapproach (87.8%). Patients with airway injuries were older (median 73 vs. 63), more likely to have squamous cell carcinoma (85.7% vs. 17.9%), andwith proximal tumors (85.7% vs. 14.1%).When given, the type of neoadjuvant treatment consisted of chemoradiotherapy in all patients who sufferedan injury, whereas it was only administered to 21.3% of patients without an injury. There were no deaths among three patients in whom the injury wasidentified intraoperatively. Mortality occurred in three of four patients (75.0%) with an injury detected postoperatively.Conclusions: Patients with proximal tumors and radiation administration as a component of neoadjuvant treatment are more likely to suffer atracheobronchial injury. An aggressive reoperative approach is warranted in patients with injuries that are discovered postoperatively.J. Surg. Oncol. 2014;109:804–807 � 2014 Wiley Periodicals, Inc.

KEY WORDS: major airway; malignancy; transhiatal; radiation

INTRODUCTION

A major airway injury during an esophagectomy is a rare, but anextremely morbid complication. The incidence ranges from 0.6% to3.9% [1–3], and the mortality can be up to 33.3% [4]. The membranousportion of the trachea is the most common location for such injuries,owing to its inherent proximity to the esophagus and relative weakness.This is particularly the case when tumors of the esophagus are in theupper or the mid‐portion of the esophagus [4]. Additionally,preoperative radiotherapy can create a fibrotic field, making anairway injury more likely [3]. An overt tracheobronchial injury willmanifest intraoperatively as a difficulty to ventilate and a presence ofsevere hypoxia. This will alert the physicians to the problem andimmediate repair can then be carried out. An insidious injury will onlycome to reveal itself in the postoperative course, and may present a moredifficult scenario to manage. Our goal was to review the outcomes ofpatients who sustained tracheobronchial injuries as a result of anesophagectomy for cancer at our tertiary referral center. We aimed tocompare this cohort of patients to those that did not have such an injury.Additionally, potential differences in perioperative outcomes weresought in patients whose injury was identified intraoperatively incomparison to a postoperative detection.

METHODS

A retrospective review was performed for patients who hadundergone esophagectomy for cancer at Jackson Memorial Hospital/University of Miami (Miami, FL) from January 2000 to June 2012. Therecords were reviewed for data on age, sex, preoperative comorbidities,tumor histology, grade, location, stage, neoadjuvant treatment and type,length of stay, complications, and mortality. Approval from InstitutionalReview Board was obtained prior to the chart review.

A total of 425 curative operations were done in this period of time.Seven patients (1.6%) sustained an injury to a major airway during thesurgery. This was identified either at the time of the operation or duringthe postoperative course. The transhiatal approach was utilized in 373patients (87.8%), Ivor‐Lewis esophagectomy in 24 patients (5.6%), anda thoracoabdominal incision in 28 patients (6.6%). These surgicalapproaches have been described in detail in the past [5–8].

Of the seven patients with an injury, six underwent neoadjuvanttreatment. Chemoradiotherapy was administered to all six patients(100%). Of the 418 patients without an injury, 300 underwentneoadjuvant treatment. Chemoradiotherapy was administered to 64(21.3%) of these patients, while chemotherapy was implemented in 236(78.7%) of 299 patients.

RESULTS

The median age for the seven patients with a tracheobronchial injurywas 73 (range 67–85), with six being male (85.7%). Majority of patientshad squamous cell carcinoma (85.7%), with poorly differentiated tumors(57.1%), and were most commonly located in the upper/mid‐esophagus(85.7%) (Table I). The median age for the 418 patients without an injury

*Correspondence to: Vadim P. Koshenkov, MD, Division of SurgicalOncology, Rutgers Cancer Institute of New Jersey, 195 Little Albany St.,New Brunswick, NJ 08901. Fax: þ1‐732‐235‐8098.E‐mail: vadim.koshenkov@rutgers.edu

Received 26 July 2013; Accepted 22 January 2014

DOI 10.1002/jso.23577

Published online 17 February 2014 in Wiley Online Library(wileyonlinelibrary.com).

Journal of Surgical Oncology 2014;109:804–807

� 2014 Wiley Periodicals, Inc.

was 63 (range 22–89), with 341 being male (81.6%). Minority of thepatients had squamous cell carcinoma (17.9%).Most patients had poorlydifferentiated tumors (40.9%) and were most commonly located in thedistal esophagus (85.9%).

Six of the seven patients underwent neoadjuvant treatment prior tothe operation for locoregionally advanced disease upon presentation,with one patient electing to proceed straight to surgery. A dose of 50Gyor higher was used in all cases, and several different chemotherapyregiments were employed. After the completion of the treatment, allpatients had a transhiatal esophagectomy. In the case of Patient 5, athoracotomy was also performed as the specimen was adherent to theaorta and the carina. The conduit had been positioned in the anteriormediastinum prior to this, and anastomosis had been created. This wasthe sole case where the gastric conduit was placed retrosternally, asopposed to the posterior mediastinum. An R0 resection wasaccomplished in six cases (85.7%).

The airway injury was detected during the operation in three patients.The identification was facilitated by a sudden difficulty with ventilation.The endotracheal tube was carefully advanced beyond the injury eitherinto the distal trachea or into a bronchus, depending on the location of theinjury. Once ventilation and oxygenation were adequate, bronchoscopywas performed in all cases with the determination of the exact location ofthe injury. For Patient 1, the cervical incision was utilized for repair asthe injury was in the upper posterior trachea, and was approximately2 cm. A primary repair with a single layer of simple interrupted 3.0 vicrylsutures was performed. For Patient 2, the location of the injury was alsoin the upper posterior trachea, and the extent was about 2.5 cm. Acervical incision was utilized, and the repair consisted of a single layer ofinterrupted vertical mattress 3.0 vicryl sutures. For Patient 7, the injurywas approximately 3 cm, and extended from above the carina into the leftmainstem bronchus posteriorly. The repair was performed via a rightthoracotomy, and consisted of a single layer of simple interrupted 4.0PDS sutures, along with a buttress of an intercostal muscle flap placedbetween the trachea and the gastric conduit.

During the postoperative course, Patient 1 developed apneumothorax, necessitating a placement of a chest tube. Patient 2developed a vent‐dependent respiratory failure (VDRF), an anastomoticleak, and a tracheogastric fistula. It was hypothesized that the leakproduced a tracheal erosion, which led to the fistula. The gastric conduitwas stented, and over time the fistula sealed. Patient 7 experienced nocomplications. The median length of stay for the three patients was36 days, with all three fully recovering.

The other four patients’ airway injuries were identified during thepostoperative period. Patient 3 developed VDRF, and on postoperativeday 28, the patient was noted to have bile coming through theendotracheal tube. Bronchoscopy showed a 2 cm injury in the posteriortrachea at the level of the carina. Also, an ischemic neoesophagus was

identified extending for 10 cm above and below the fistula site. Twoendotracheal tubes were advanced beyond the fistula and adequateventilation was achieved. However, over the next 4 days, the patientdeveloped multiple organ dysfunction, had the status changed to DNR/DNI after a discussion with the family, and passed away.

Patient 4 had a steady‐postoperative course until a reintubation wasrequired for a respiratory decompensation on postoperative Day 18.Bronchoscopy for the purposes of clearing the airways showed a distalposterior tracheal injury, which was approximately 1 cm. Afterwards,subcutaneous emphysema was noted along the chest wall and the neck.Tracheal stenting was performed, and this stabilized the respiratorystatus for the following several days. The patient continued to havesubcutaneous emphysema, albeit diminished from before the stentingprocedure. However, 2 days after the stent was placed, the patientexperienced a sudden significant hemorrhage from the cervical incision,became asystolic, and could not be resuscitated.

Patient 5 required multiple reintubations for respiratory failure. Abronchoscopy on postoperative Day 21 revealed a 1–2mm defect in theposterior left mainstem bronchus. The neoesophagus was in closeapposition to the injury, and so it was managed expectantly. After aprolonged hospital stay of 59 days, the patient recovered fully.

Patient 6 was explored twice in the first 24 hr after the operation forcontrol of postoperative hemorrhage that was from a raw surface of theposterior mediastinum. Subsequently, the patient developed VDRF. Onpostoperative Day 24, at the time of an attempted tracheostomy, abronchoscopy revealed a high posterior tracheal injury of 2 cm locatedapproximately 6 cm above the carina. The tracheostomy was aborted forconcerns of enlarging the airway defect with manipulation. An upperendoscopy, the same day demonstrated a tracheogastric fistula and aviable neoesophagus. The use of endoclips was unsuccessful in closingthe defect of the neoesophagus. The endotracheal tube was advancedinto the distal trachea, but difficulties with the respiratory statuspersisted. The fistula could not be sealed, and it was felt that the injuryhad extended distally. Over the next 3 days, the patient’s hemodynamicstatus deteriorated, there was an asystolic episode, and he could not beresuscitated. Of the four patients who had postoperative detection of anairway injury, three did not survive (75.0%).

DISCUSSION

In our cohort, a tracheobronchial injury was encountered in sevenpatients, at a rate of 1.6%. The incidence of this complication is withinthe range of the previously reported data [1–3,9]. The seven patientsdiffered considerably from the patients without an airway injury.Squamous cell carcinoma, proximal andmid‐esophageal tumor location,and neoadjuvant chemoradiotherapy as opposed to adenocarcinoma,distal tumor location, and neoadjuvant chemotherapy were more

TABLE I. Patient Demographics and Outcomes

Factor Patient 1 Patient 2 Patient 3 Patient 4 Patient 5 Patient 6 Patient 7

Age 68 72 73 67 79 85 74Sex Male Male Male Male Female Male MaleHistology Squamous Squamous Squamous Adeno Squamous Squamous SquamousLocation of tumor Upper Upper Upper Mid Mid Lower MidNEO CRT No Yes Yes Yes Yes Yes YesIntraop Yes Yes No No No No YesSite of airway injury Proximal

tracheaProximaltrachea

Distaltrachea

Distaltrachea

Leftmainstembronchus

Proximaltrachea

Distaltrachea/Lbronchus

Size (cm) 2.0 2.5 2.0 1.0 0.2 2.0 3.0LOS (days) 36 136 32 22 59 28 13Mortality No No Yes Yes No Yes no

NEO, neoadjuvant; CRT, chemoradiotherapy; LOS, length of stay; Adeno, adenocarcinoma.

Journal of Surgical Oncology

Tracheobronchial Injury in Esophagectomy 805

common in this group of patients. These differences underscore theimportance of approaching such patients with extra caution at the time ofthe operation. Even though the incidence of a major airway injury in thesetting of an esophagectomy is quite low, in patients with proximallesions that are treated with preoperative radiation, it is not insignificant.Onemust apply extra care to the blunt dissection of the upper esophagus,and ensure that visualization is optimal [4]. The exposure can beimproved with a phrenotomy or a resection of the medial end of theclavicle and part of the manubrium.We did not employ these techniquesat any of the operations, as it was felt that visualization was optimal.However, the surgeon should have a lower threshold to perform thesemaneuvers in cases with a higher risk of a tracheobronchial injury.Another way to potentially optimize exposure would be to perform athoracotomy, but the incidence of a major airway injury withtransthoracic esophagectomy has been reported to be 1.5–6.0% [3,9,10], which is not lower than for the transhiatal approach.In light of these numbers, a transthoracic approach does not appear tohave a lower incidence of a tracheobronchial injury.

It is worth commenting on the likely etiologies of the airway injuriessustained in our cohort. In general, such injuries occur as the result ofintraoperative trauma, anastomotic leaks that lead to an inflammatoryprocess near the tracheobronchial tree, pressure from a cuff of anendotracheal tube in cases of prolonged intubation, or ischemia fromextensive dissection in the upper mediastinum [3]. It is clear that directinjury to the membranous trachea occurred in the three cases that werediscovered intraoperatively. Two of the postoperatively detectedlesions most likely resulted from ischemia. Bartels et al. [3] closelyexamined 31 patients with tracheobronchial lesions. They determinedthat injuries of ischemic origin were usually located around the carina,presented 7–17 days after the resection andwere the most frequent ones.Lesions from etiologies such as surgical injury that was missedat the time of the operation, cuff pressure‐related, positive‐pressureventilation in patients with VDRF, and inflammation from ananastomotic leak were all located above the carina. Moreover, theseinjuries were discovered either within a week of the operation or after17 days, depending on the etiology. We believe that Patient 4 had anischemic injury as there was no evidence of an anastomotic leak, noprolonged period of intubation, and the injury was pericarinal. Patient 5had an extensive posterior mediastinal dissection as the tumor wasadherent to the aorta and the tracheobronchial tree. This was the onlypatient who required a thoracotomy to remove the specimen. For thisreason and since there was no anastomotic leak, we believe the injurywas ischemic in nature. Patients 3 and 6 developed tracheogastricfistulas, andmost likely, an anastomotic leak occurred first and led to thedevelopment of the fistula. Buskens et al. [11] reported that ananastomotic leak preceded the tracheogastric fistula in all six patients oftheir cohort. Similarly, Yasuda et al. [10] found that anastomoticleakage was responsible for the majority of the 10 gastro‐tracheobronchial fistulas. In some patients, gastric conduit necrosisled to the airway injury, and this remains a possibility for Patient 3 in ourcohort, as there was evidence of an ischemic neoesophagus at the timeof the bronchoscopy.

Despite postoperative complications, some severe, all three patientswith an intraoperative repair of a tracheobronchial injury made a fullrecovery. This is consistent with previous reports [12,13]. The decisionto suture the defect closed was based on the size (2 cm or larger), and thisis in agreement with a report by Gupta et al. [13], who only repairedlacerations that were 3 cm or longer. These authors felt that for injuriesless than 2 cm, gastric tube reinforcement alone was sufficient to heal thelaceration, and their patients that were managed in this fashion recoveredfully. In this method, the gastric conduit acts like a patch and isolates thetear from the environment. For two of our patients, the cervical incisionwas utilized to repair injuries that were proximal to the carina. We werehesitant to perform a repair through the cervical incision in the patientwith the laceration extending into the left mainstem bronchus. For this

reason, a right thoracotomy was employed, and a local muscle flap wasused as a buttress in addition to the primary repair. In contrast, Hulscheret al. [12] used a right thoracotomy in all five patients that wereundergoing a transhiatal esophagectomy. This is despite the fact that allinjuries were proximal to the carina. Interestingly, Gupta et al. [13] weresuccessful in repairing a 5 cm laceration that extended into the rightmainstem bronchus via the cervical incision. The use of a headlight andwell‐positioned retractors, one on the trachea and another on thesternocleidomastoid muscle, allowed for a good exposure of the retro‐tracheal space.

The discovery of a tracheobronchial injury during the postoperativecourse was under different circumstances in all four of our patients. Inone of the patients with a tracheogastric fistula, bile was noticed in theendotracheal tube. In the other, the lesion was identified at the time of anattempted tracheostomy. The lesions were detected in the other twopatients during a bronchoscopy after respiratory decompensation hadoccurred. It has been reported that the presence of a persistent air leak orsubcutaneous emphysema can alert the surgeon to a possibletracheobronchial injury [4,12]. Unfortunately, only Patient 6 had apersistent air leak with full lung expansion, and none of the four patientsshowed signs of subcutaneous emphysema prior to the identification ofthe injury. This did not allow for an early detection of the injuries.

Nonoperative management was attempted in all four patients. It wasfelt that the small defect in the left mainstem bronchus of Patient 5 wouldheal spontaneously as the gastric conduit was abutting the site of theinjury. The other three patients were considered to be poor surgicalcandidates for a reoperation because of advanced chronic obstructivepulmonary disease. For this reason, different nonoperative strategieswere employed in order to manage the airway injuries. The advancementof the endotracheal tube beyond the laceration while maintainingpositive‐pressure ventilation [14], tracheal stenting [3,15], andendoscopic obliteration [16] have been described in the literature.However, the success of these interventions depends on several factors.Multiple authors have suggested that an attempt at nonoperativemanagement may be warranted when the lung is fully expanded,oxygenation is unimpaired, and there is no mediastinal or pulmonaryinfection [3,4,9,11,12]. In the presence of severe symptoms, aggressivesurgical intervention whenever possible is central to the management ofthis grave complication. Even with a reoperation, the mortality has beenreported to be very high at around 30% [3,4,10].

In conclusion, tracheobronchial lesions are not as rare in the subgroupof patients who have proximal lesions and undergo radiation as part ofneoadjuvant therapy, as they are in all patients who undergo anesophagectomy for cancer. When detected intraoperatively during atranshiatal approach, immediate repair can be performed through thecervical incision as long as the exposure is adequate, and can be aided byproper surgeon positioning and careful retraction. These patients tend torecover fully from their injuries. A major airway injury that is identifiedduring the postoperative period can be very difficult to manage. Earlyrecognition and spontaneous breathing will aide in the resolution of theinjury. Prompt surgical intervention should not be delayed for patientswith severe symptoms, such as a persistent air leak, poor oxygenation,and infection. Nonoperative measures will likely result in a highmortality rate in such patients.

REFERENCES

1. Orringer MB, Marshall B, Chang AC, et al.: Two thousandtranshiatal esophagectomies: Changing trends, lessons learned. AnnSurg 2007;246:363–372.

2. Hulscher JB, Tijssen JG, Obertop H, et al.: Transthoracic versustranshiatal resection for carcinoma of the esophagus: A meta‐analysis. Ann Thorac Surg 2001;72:306–313.

3. Bartels HE, Stein HJ, Siewert JR: Tracheobronchial lesionsfollowing oesophagectomy: Prevalence, predisposing factors andoutcome. Br J Surg 1998;85:403–406.

Journal of Surgical Oncology

806 Koshenkov et al.

4. Kannan RR, Mahajan V: Tracheal injury during transhiatalmobilization of the esophagus. J Surg Oncol 1999;71:186–188.

5. Orringer MB, Sloan H: Esophagectomy without thoracotomy. JThorac Cardiovasc Surg 1978;76:643–654.

6. OrringerMB,Marshall IannettoniMD: Transhiatal esophagectomy:Clinical experience and refinements. Ann Surg 1999;230:392–400.

7. Lewis I: The surgical treatment of carcinoma of the esophagus withspecial reference to a new operation for growths of the middle third.Br J Surg 1946;34:18–31.

8. Heitmiller RF: Results of standard left thoracoabdominal esoph-agogastrectomy. Semin Thorac Cardiovasc Surg 1992;4:314–319.

9. Maruyama K, Motoyama S, Sato Y, et al.: Tracheobronchial lesionsfollowing esophagectomy: Erosions, ulcers, and fistulae, and thepredictive value of lymph node‐related factors. World J Surg2009;33:778–784.

10. Yasuda T, Sugimura K, Yamasaki M, et al.: Ten cases of gastro‐tracheobronchial fistula: A serious complication after esophagec-tomy and reconstruction using posterior mediastinal gastric tube.Dis Esophagus 2012;25:687–693.

11. Buskens CJ, Hulcher JB, Fockens P, et al.: Benign trachea‐neo‐esophageal fistulas after subtotal esophagectomy. Ann Thorac Surg2001;72:221–224.

12. Hulscher JB, ter Hofstede E, Kloek J, et al.: Injury to themajor airways during subtotal esophagectomy: Incidence, manage-ment, and sequelae. J Thorac Cardiovasc Surg 2000;120:1093–1096.

13. Gupta V, Gupta R, Thingnam SK, et al.: Major airway injury duringesophagectomy: Experience at a tertiary care center. J GastrointestSurg 2009;13:438–441.

14. Wylezol M, Wiewiora M, Pardela M: Tracheal injury duringtranshiatal esophagectomy treated conservatively. Wiad Lek2004;57:180–182.

15. Freitag L, Tekolf E, Stamatis G, et al.: Clinical evaluation of a newbifurcated dynamic airway stent: A 5‐year experience with 135patients. Thorac Cardiovasc Surg 1997;45:6–12.

16. Marone G, Santoro LM, Torre V: Successful endoscopic treatmentof GI‐tract fistulas with a fast‐hardening amino acid solution.Endoscopy 1989;21:47–49.

Journal of Surgical Oncology

Tracheobronchial Injury in Esophagectomy 807