ARTICLE IN PRESS

www.icvts.org

doi:10.1510/icvts.2010.247866

Interactive CardioVascular and Thoracic Surgery 12 (2011) 147–151

� 2011 Published by European Association for Cardio-Thoracic Surgery

New

IdeasInstitutional

ReportW

orkin

ProgressReport

ESCVSArticle

NegativeResults

State-of-the-artBest

EvidenceTopic

BriefCom

munication

CaseReport

Follow-up

PaperEditorial

ProtocolProposalfor

Bail-out

ProcedureN

omenclature

HistoricalPages

Institutional report - Esophagus

Treatment of intrathoracic esophageal anastomotic leaks by meansof endoscopic stent implantation�

Michael Schweigert *, Attila Dubecz , Rudolf J. Stadlhuber , Herbert Muschweck , Hubert J. Steina, a a b a

Department of General and Thoracic Surgery, Klinikum Nuernberg Nord, Nuernberg, Germanya

Department of Gastroenterology, Klinikum Nuernberg Nord, Nuernberg, Germanyb

Received 3 July 2010; received in revised form 2 November 2010; accepted 4 November 2010

Abstract

Intrathoracic anastomotic leakage in patients with esophagectomy is associated with high morbidity and mortality. Until recently surgicalreexploration was the preferred way of dealing with this life-threatening complication. But mortality remained significant. After the firstsuccessful reports we adopted endoscopic stent implantation as a primary treatment option. The aim of this study was to investigate thefeasibility and the results of endoscopic stent implantation. Between January 2004 and December 2009, 167 patients underwent anesophageal resection. Surgery was mainly the result of esophageal cancer. An intrathoracic esophageal anastomotic leak was endoscopicallyverified in 17 patients. Twelve patients received an implantation of a self-expanding stent as a primary treatment. An endoscopic stentplacement was accomplished in all 12 patients. In nine patients a definitive closure of the leak was achieved and the stent couldsubsequently be removed. Two patients died due to severe sepsis in spite of sufficient stent placement. Because of early recurrence ofvery malign small cell cancer the stent remained in situ in one patient. In conclusion, stent implantation for intrathoracic esophagealanastomotic leaks is feasible and compares favorable with the results of surgical reexploration. It is an easily available minimally-invasiveprocedure which may reduce leak-related mortality and morbidity.� 2011 Published by European Association for Cardio-Thoracic Surgery. All rights reserved.

Keywords: Esophageal cancer; Esophagectomy; Intrathoracic anastomotic leakage; Self-expanding stent; Endoscopic stent implantation; Sepsis

1. Introduction

In the past two decades the outcome in esophagealsurgery has constantly improved w1, 2x. In specializedesophageal surgery units morbidity and mortality have beenconsiderably reduced w1x. Notwithstanding this favorabletrend anastomotic leakage remains a major source of deathafter esophagectomy w3x. Particularly intrathoracic anasto-motic leakage is associated with high morbidity and mor-tality. The lethality of this devastating complication is upto 40% w3–6x. Until recently surgical reexploration was thepreferred way of dealing with this life-threatening compli-cation. But the results have not been convincing and insome studies surgical treatment of esophageal anastomoticleaks has led to high mortality w4x. While sepsis sourcecontrol, sufficient drainage of the leakage and preventionof further mediastinal contamination are generally accept-ed aims of the immediate treatment the way to achievethese goals remain controversial w3x. Conservative treat-ment of minor leaks became an established treatmentalternative w7x and in the recent years we have also seen

� Presented at the 18th European Conference on General Thoracic Surgery,Valladolid, Spain, May 30–June 2, 2010.

*Corresponding author. Klinik fuer Allgemein-, Viszeral- und Thoraxchirur-gie, Klinikum Nuernberg Nord, Prof-Ernst-Nathan-Str. 1, 90419 Nuernberg,Germany. Tel.: q49-911-3983566/2979; fax: q49-911-3982193.

E-mail address: michael.schweigert@klinikum-nuernberg.de(M. Schweigert).

the first reports of successful endoscopic stent insertion inpatients with intrathoracic esophageal anastomotic leaksw8x. Therefore, we adopted an endoscopic placement of aself-expanding metal stent as a primary treatment optionin patients with endoscopically verified intrathoracic anas-tomotic leaks if there was no ischemia or necrosis of thepulled up gastric tube. The aim of this study was toinvestigate the feasibility and the results of endoscopicstent implantation.

2. Patients and methods

2.1. Patients

The study includes 167 consecutive patients who under-went an esophageal resection between January 2004 andDecember 2009 at the Department of General and ThoracicSurgery of the Klinikum Nuernberg Nord. The median agewas 61.8 years and there were 142 male (85%) and 25female (15%) patients.

The main reason for surgery was esophageal cancer: therewere 73 squamous cell carcinomas (43.7%), 66 adenocar-cinomas of the esophagogastric junction (AEG) type I(39.5%), 19 AEG type II (11.4%), which is a cancer of thegastric cardia with involvement of the distal esophagus w9–11x, four other malignant tumors (2.4%) and five benignlesions (3.0%) (Table 1).

ARTICLE IN PRESS

148 M. Schweigert et al. / Interactive CardioVascular and Thoracic Surgery 12 (2011) 147–151

Table 1. Histological classification of tumors

Histology Total number %

Squamos cell carcinoma 73 43.7Adenocarcinoma of the esophagogastric junction type I (adenocarcinoma of the distal esophagus) 66 39.5Adenocarcinoma of the esophagogastric junction type II (adenocarcinoma of the gastric cardia with involvement of the esophagus) 19 11.4Neuroendocrine esophageal cancer 2 1.2Adenosquamos carcinoma 1 0.6Leiomyosarcoma of the esophagus 1 0.6Benign lesions 5 3.0

Fig. 1. Radioscopic image of an inserted self-expanding metal stent. A chesttube provides drainage of the right pleural cavity.

Fig. 2. CT of the chest in the same patient shows the successful closure ofthe anastomotic leakage. The swallowed contrast medium remains within thestent which proofs the sealing of the leak. Also the chest tube can be seennear the insufficiency. CT, computed tomography.

A total of 39 patients (23.4%) received neoadjuvant ther-apy, either neoadjuvant chemoradiotherapy in the presenceof squamous cell carcinoma or neoadjuvant chemotherapyalone in the case of adenocarcinoma of the distal esophagusor the gastric cardia (AEG type I and II).

In 125 patients an abdomino-thoracic esophagectomy withgastric pull up and intrathoracic-stapled anastomosis wasperformed (74.9%). Furthermore, four transmediastinalesophagectomies (2.4%), 15 extended total gastrectomieswith transhiatal resection of the distal esophagus andintrathoracic-stapled esophago-jejunostomy (9%), 12 trans-thoracic esophagectomies with delayed reconstruction(7.2%) and 11 other procedures (6.5%) were carried out.All patients routinely received thoracic epidural analgesiafor improved postoperative pain control.

2.2. Diagnosis of anastomotic leakage

An intrathoracic esophageal anastomotic leak was endo-scopically verified in 17 patients. If leakage of the intra-thoracic anastomosis was clinically suspected it was alwaysidentified by endoscopy and computed tomography (CT) ofthe chest and abdomen. Only endoscopically verified leak-ages are part of this study.

2.3. Stent insertion

A self-expanding, covered metal stent (Ultraflex�, BostonScientific�, Natick, MA, USA) was used to close the anas-tomotic leak. The stent placement was performed by agastroenterologist well trained in interventional endoscopy.The exact position of the leakage was marked on thepatients skin and afterwards the stent was inserted underradioscopical guidance. After implantation the correctplacement of the stent and the successful closure ofthe leak were always endoscopically and radioscopicallychecked (Figs 1–4).

3. Results

In 12 out of the 17 endoscopically verified intrathoracicanastomotic leaks endoscopic implantation of a self-expanding metal stent was successfully accomplished (Table2). Rethoracotomy was mandatory in the five other casesbecause of either necrosis and ischemia of the pulled upgastric tube or advanced pleural empyema that requiredsurgical debridement.

After stent placement contrast swallow esophagographyand endoscopy showed a complete sealing of the anasto-motic leak (Figs. 1–4). Stent migration occurred but endo-scopic reintervention was always feasible.

All patients were treated at the intensive care unit andreceived sepsis therapy including antibiotherapy, hemodyn-

ARTICLE IN PRESS

149M. Schweigert et al. / Interactive CardioVascular and Thoracic Surgery 12 (2011) 147–151

New

IdeasInstitutional

ReportW

orkin

ProgressReport

ESCVSArticle

NegativeResults

State-of-the-artBest

EvidenceTopic

BriefCom

munication

CaseReport

Follow-up

PaperEditorial

ProtocolProposalfor

Bail-out

ProcedureN

omenclature

HistoricalPages

Fig. 3. Radioscopic image of an inserted self-expanding metal stent. Alsovisible is a nasogastric tube for decompression of the pulled up gastric tube.

Fig. 4. CT-scan of the same patient. The stent closed the leak successful andwas well placed.

Table 2. Characteristics of patients with stent implantation

Number of patients treated with endoscopic stent 12implantationMedian age 59.5 yearsSquamos cell carcinoma 2Adenocarcinoma of the esophagogastric junction type I 8Neuroendocrine esophageal cancer 2Neoadjuvant chemoradiotherapy 1Stents explanted 10Median time between stent implantation and final 48.4 daysexplantation (16–99 days)

amic monitoring and management, hemofiltration if nec-essary and adjunctive therapies. Physiotherapy and inhala-tion with positive airway pressure were administered assoon as possible. Tube thoracostomy was performed in allpatients and some patients received irrigation of the pleu-ral cavity with Ringer solution via the inserted chest tubes.

Besides stent dislocation, which could always be correctedendoscopically, one severe and lethal complication occur-red. One patient developed a stent related erosion of theaorta thoracica 17 days after stent insertion which led toa fistula between the aorta and the esophagus. The patientdied of sudden massive hemorrhage. The aortic erosion andthe consecutive fistula were confirmed by autopsy.

Stent removal was performed in 10 patients routinely.After stent explantation the underlying mucosa showedneither necrosis nor severe morphological changes. How-ever, mucosal hyperproliferation was observed at the stentmargins. In nine patients endoscopy showed no sign of leakpersistence. In one case after stent removal a tiny fistulawith a diameter of c. 3 mm was detected and successfullysealed with cyanoacrylate glue. In spite of accomplishedhealing of the anastomotic leakage one patient died

24 days after successful stent explantation because ofmassive pulmonary aspiration.

In one case the stent remained in situ. The 47-year-oldpatient suffered from a extremely malignant G4 neuro-endocrine and small cell cancer of the esophagus. Becauseof anastomotic leakage a stent was inserted and closure ofthe leak was successfully achieved. The patient was thentransferred to a rehabilitation clinic. About three monthslater he complained about pain and a cancer recurrence inthe mediastinum was diagnosed. He died six months afterthe initial esophagectomy because of rapid progressingcancer with local recurrence and distant metastasis.

The in-hospital mortality was 16.7% (2y12), in 10 patients(83.3%) successful closure of the anastomotic leakage wasachieved and the patients were transferred to rehabilita-tion facilities and finally discharged home.

4. Discussion

Advances in surgical technique and perioperative careincluding intensive care and anesthesia has improved theoutcome in esophageal surgery in the last decades. Thishas been attributed to the development of specialized

ARTICLE IN PRESS

150 M. Schweigert et al. / Interactive CardioVascular and Thoracic Surgery 12 (2011) 147–151

esophageal surgery units with greater experience, selectionof patients and refinement and standardization of surgicalprocedures and perioperative management w1, 2x. The in-hospital mortality after esophageal resection has decreasedfrom 29% to 7.5% in the period between 1950 and 2000 w1,2, 7x. Notwithstanding this favorable trend, anastomoticleakage is still the most common severe postoperative com-plication w3, 7x. Intrathoracic anastomotic leaks accountfor c. 40% of postoperative deaths after esophagectomyw6x. The fatality of this crushing complication is up to 40%w3–6x. If proper drainage of the anastomotic leakage is notachievable the fatality reaches 80% w3x. Surgical reexplor-ation is associated with a high mortality between 60% andeven 100% w4, 12x. Because of this unfavorable resultconservative treatment options with drainage, total par-enteral nutrition, nasogastric decompression and antibio-therapy have been suggested w6, 7x. The problem with thisconservative treatment is the continued pollution of themediastinum and the pleural cavity through the anasto-motic leak. This accounts for mediastinitis, pleural empy-ema and sepsis. The mortality rate of this conservativetreatment has been similar to surgical reexploration andup to 40%. For this reason conservative treatment withoutendoscopic closure of the leak is nowadays only acceptedin small anastomotic dehiscences which are well drainedw3, 6x.

Therefore, adequate external drainage of the leak as wellas prevention of further mediastinal contamination are theprimary therapeutic goals in the management of intrathor-acic anastomotic leakage w3x. While sufficient drainage canbe achieved by tube thoracostomy or interventional drain-age the best way to prevent further pollution and contam-ination of the mediastinum remains controversial.

Several endoscopic devices have been suggested for clo-sure of anastomotic leaks. For example endoscopic clippingor endoscopic sealing with cyanoacrylate glue w8, 13–15x.However, all these techniques are again only feasible andpractical in small leaks and so the problem of largerdehiscences remains unsolved.

Self-expanding endoscopic stents have been successfullyused in the treatment of inoperable esophageal obstruc-tions. The first reports of endoscopic stent implantation inthe treatment of intrathoracic anastomotic leaks afteresophagectomy have been published w8x. Self-expanding,covered metal stent (e.g. Ultraflex�) can close leaksregardless of size as long as the pulled up gastric tube isnot ischemic and there is no complete dehiscence of theanastomosis.

As a result, in the Department of General and ThoracicSurgery at the Klinikum Nuernberg we changed our treat-ment approach for anastomotic leaks in January 2004 andadopted endoscopic stent insertion as the primary treat-ment option. The aim of this study was to investigate thefeasibility and the results of endoscopic stent implantation.The study includes 167 consecutive patients who underwentesophageal resection between January 2004 and December2009.

The rate of postoperative anastomotic leakages was 10%(17y167). This is comparable to other series, where in thelast two decades anastomotic leakage rates between 4%and 11% have been reported w13x.

If leakage of the intrathoracic anastomosis was clinicallysuspected it was always identified by endoscopy and CT ofthe chest and abdomen. All 17 anastomotic leaks wereendoscopically verified.

The convincing advantage of endoscopy over radiographiccontrast-medium swallow in the diagnosis of intrathoracicanastomotic leakage is the possibility of direct visual exam-ination of the anastomosis, quantification of the leak anddetermination whether the pulled up gastric tube is ische-mic or not w3, 8x. Therefore, an immediate decision wheth-er endoscopic stent insertion is feasible or not can be madeout of the endoscopic inspection of the leakage. In additionto the endoscopy a CT of the chest and abdomen ismandatory to rule out advanced pleural empyema or medi-astinal abscess which would require either percutaneousinterventional or surgical drainage.

In 12 patients an endoscopic stent insertion was success-fully accomplished. Immediate closure of the leak wasachieved in all 12 patients. In five patients either becauseof necrosis of the pulled up gastric tube or because ofadvanced pleural empyema and subsequent severe sepsisor even septic shock a rethoractomy was mandatory. Insuch cases a surgical reexploration with take down of theanastomosis, resection of the ischemic gastric tube as wellas debridement and irrigation of the pleural cavity areunavoidable w5, 6, 12x.

There was one devastating stent-related complication.One patient developed an erosion of the aorta thoracicawhich led to a fistula between the aorta and the esophagus.The patient died due to sudden massive hemorrhage. How-ever, such septic vascular erosions are a rare but notunknown complication of anastomotic leaks with consecu-tive contamination of the mediastinum w12x. They are notrestricted to patients with implanted stents but can beseen in all patients with intrathoracic anastomotic leaksindependent of their initial treatment. Nevertheless stentsmight be an additional risk for vascular erosion because ofthe mechanical alteration of the surrounding tissue.

Another patient died because of severe pulmonary aspi-ration 24 days after successful stent explantation. The totalin-hospital mortality was 16.7% (2y12).

This in-hospital mortality of only 16.7% is comparable withreports of other groups, that describe a mortality of c.25%, and is far better than the outcome of conservativetherapy without a stent, where a fatal outcome in up to40% of patients is to be expected.

The anastomotic leaks of 83.3% of the patients (10y12)were successfully managed by stent insertion. All thesepatients were transferred first to a rehabilitation facilityand were finally discharged from hospital. Out of these 10patients nine had an excellent outcome. The stents wereremoved in these patients and proper oral food intake waspossible. Stent removal could be performed without com-plications. One younger patient (47 years) with an excep-tional malign neuroendocrine and small cell carcinoma ofthe esophagus suffered from an early cancer recurrence inthe mediastinum with involvement of the trachea. Hence,the stent remained in situ to prevent a esophago-trachealfistula and to secure the oral ingestion of food. The patientdied c. six months after initial esophagectomy due to localcancer progress and distant metastasis. However, this death

ARTICLE IN PRESS

151M. Schweigert et al. / Interactive CardioVascular and Thoracic Surgery 12 (2011) 147–151

New

IdeasInstitutional

ReportW

orkin

ProgressReport

ESCVSArticle

NegativeResults

State-of-the-artBest

EvidenceTopic

BriefCom

munication

CaseReport

Follow-up

PaperEditorial

ProtocolProposalfor

Bail-out

ProcedureN

omenclature

HistoricalPages

was not related to the anastomotic leak but represents theextreme malignity of the underlying disease.

In conclusion, this outcome study shows that successfultreatment of intrathoracic anastomotic leaks after esopha-gectomy by means of endoscopic stent insertion is feasiblewith good results regarding closure of the leak and recoveryof the patients.

References

w1x Bonavina L, van Lanschot JJB. Complications in oesophageal and gastricsurgery. Introduction. Dig Surg 2002;19:86–87.

w2x Abunasra H, Lewis S, Beggs L, Duffy J, Beggs D, Morgan E. Predictorsof operative death after oesophagectomy for carcinoma. Br J Surg 2005;92:1029–1033.

w3x Siewert JR, Stein HJ, Bartels H. Insuffizienzen nach Anastomosen imBereich des oberen Gastrointenstinaltraktes. Chirurg 2004;75:1063–1070.

w4x Alanezi K, Urschel JD. Mortality secondary to esophageal anastomoticleak. Ann Thorac Cardiovasc Surg 2004;10:71–75.

w5x Griffin SM, Lamb PJ, Dresner SM, Richardson DL, Hayes N. Diagnosisand management of mediastinal leak following radical oesophagectomy.Br J Surg 2001;88:1346–1351.

w6x Sauvanet A, Baltar J, Le Mee J, Belghiti J. Diagnosis and conservativemanagement of intrathoracic leakage after oesophagectomy. Br J Surg1998;85:1446–1449.

w7x Lerut T, Coosemans W, Decker G, De Leyn P, Nafteux P, Van RaemdonckD. Anastomotic complications after esophagectomy. Dig Surg 2002;19:92–98.

w8x Kauer WK, Stein HJ, Dittler HJ, Siewert JR. Stent implantation as atreatment option in patients with thoracic anastomotic leaks afteresophagectomy. Surg Endosc 2008;22:50–53.

w9x Stein HJ, Feith M, Siewert JR. Cancer of the esophagogastric junction.Surg Oncol 2000;9:35–41.

w10x Stein HJ, Feith M, Siewert JR. Individualized surgical strategies forcancer of the esophagogastric junction. Ann Chir Gynaecol 2000;89:191–198.

w11x Siewert JR, Stein HJ, Sendler A, Fink U. Surgical resection for cancerof the cardia. Semin Surg Oncol 1999;17:125–131.

w12x Urschel JD. Esophagogastrostomy anastomotic leaks complicating eso-phagectomy: a review. Am J Surg 1995;169:634–640.

w13x Holscher AH, Schroder W, Bollschweiler E, Beckurts KTE, Schneider PM.Wie sicher ist die hoch intrathorakale Osophagogastrostomie? Chirurg2003;74:726–733.

w14x Pross M, Manger T, Reinheckel T, Mirow L, Kunz D, Lippert H. Endoscopictreatment of clinically symptomatic leaks of thoracic esophageal anas-tomoses. Gastrointest Endosc 2000;51:73–76.

w15x Rodella L, Laterza E, De Manzoni G, Kind R, Lombardo F, Catalano F,Ricci F, Cordiano C. Endoscopic clipping of anastomotic leakages inesophagogastric surgery. Endoscopy 1998;30:453–456.

eComment: Treatment of thoracic anastomotic leaks afteresophagectomy

Authors: Stefano Cafarotti, Department of Thoracic Surgery, CatholicUniversity, 00168 Rome, Italy; Filippo Lococo, Maria Letizia Vita, VenanzioPorziella

doi:10.1510/icvts.2010.247866AWe have read with interest the report by Michael Schweigert and col-

leagues on self-expanding stent as primary treatment of intrathoracic leakafter esophagectomy w1x.

Controversies still exist regarding the best treatment in cases of post-esophagectomy anastomotic leak. When this complication occurs, the relat-ed mortality rate can reach more than 60% also due to lack of standardized

treatment algorithm w2x. The comprehensive classification from Lerut andco-workers w3x provides a good stratification in the case of esophagealfistulas.

We agree with the authors that the use of modern endoscopic techniqueswould potentially reduce morbidity but we consider this into a multidiscipli-nary therapeutic algorithm based on a specific classification of the fistulathat can be resumed as I) subclinical (drainage passage of methylene bluesolution, without a clear radiological confirmation); II) minor (radiologicallyproven fistula with minor clinical complications); III) major (radiologicallyproven fistula with major clinical complications); IV) complete (gastricnecrosis). Therefore, we generally use the self-expanding stents as primarychoice only in the treatment of major intrathoracic leak or after unsuccessfulconservative treatment of cervical anastomotic leaks in transhiatal esopha-gectomy. According to encouraging results in the closure of intrathoracicanastomotic leakage reported by the authors, it would be very useful tokeep in mind the definite size of the ‘leak’ in the selection criteria for stentplacement, this crucial data being not mentioned by the authors themselves.

Finally, although the obvious benefits of self-expanding prosthesis inintrathoracic fistulas, we believe that transhiatal esophagectomy, whentechnically feasible, could represent the safest procedure to substantiallydecrease the mortality rate of anastomotic leaks w4x.

References

w1x Schweigert M, Dubecz A, Stadlhuber RJ, Muschweck H, Stein HJ.Treatment of intrathoracic esophageal anastomotic leaks by means ofendoscopic stent implantation. Interact CardioVasc Thorac Surg2011;12:147–151.

w2x Whooley BP, Law S, Alexandrou A, Murthy SC, Wong J. Critical appraisalof the significance of intrathoracic anastomotic leakage after esopha-gectomy for cancer. Am J Surg 2001;181:198–203.

w3x Lerut T, Coosemans W, Decker G, De Leyn P, Nafteux P, van RaemdonckD. Anastomotic complications after esophagectomy. Dig Surg 2002;19;92–98.

w4x Cafarotti S, Cesario A, Porziella V, Granone P. Intrathoracic manifesta-tions of cervical anastomotic leaks after transhiatal and transthoracicoesophagectomy. Br J Surg 2010;97:726–731.

eComment: Intrathoracic esophagogastric anastomotic leakagefollowing esophageal surgery

Authors: Nikolaos Barbetakis, Department of Thoracic Surgery, Theagen-io Cancer Hospital, Al. Simeonidi 2, Thessaloniki, Greece; Christos Asteriou,Athanassios Kleontas, Christodoulos Tsilikas

doi:10.1510/icvts.2010.247866BWe have read with great interest the article by Schweigert et al. concern-

ing the treatment of intrathoracic esophageal anastomotic leaks by meansof endoscopic stent implantation w1x.

An important issue which is not clarified by the authors, is the relationbetween the fistula size and success of the proposed method, as well as theoptimal time for stent removal.

We also have three points for interactive discussion. The first point is thatsuccessful sealing of the leak is gratifying and important but it is not thecomplete management of an esophageal leakage. Other important compo-nents are the drainage of extraluminal fluid collections, decortication forpleural sepsis and separation of esophagus from adjacent structures, suchas the aorta or airway, with vascularized tissue making the choice of theappropriate method of management difficult and debatable. The secondpoint is how much we trust the scar that results from the stent alone.Another issue we would like to highlight is the reinforcement of esophago-gastric anastomosis during surgery with vascularized tissue (pleural tentingor muscular flap) in order to prevent leakage.

Reference

w1x Schweigert M, Dubecz A, Stadlhuber RJ, Muschweck H, Stein HJ.Treatment of intrathoracic esophageal anastomotic leaks by means ofendoscopic stent implantation. Interact CardioVasc Thorac Surg 2011;12:147–151.