Original article

Impact of endoscopic assessment and treatment on operative and non-operative management of acute oesophageal perforation
M. K. Kuppusamy1 , C. Felisky1 , R. A. Kozarek2 , D. Schembre2 , A. Ross2 , I. Gan2 , S. Irani2 and D. E. Low1
Departments of 1 Surgery and 2 Gastroenterology, Virginia Mason Medical Center, Seattle, Washington, USA Correspondence to: Dr D. E. Low, Department of Surgery, Virginia Mason Medical Center, 1100 Ninth Avenue C6-GSUR, Seattle, Washington 98111, USA (e-mail: gtsdel@vmmc.org)

Background: Surgeons have not typically utilized an endoscopic approach for diagnosis and management of acute oesophageal perforation, mainly due to fears of increased mediastinal contamination. This study assessed the evolution of endoscopic approaches and their effect on outcomes over time in acute oesophageal perforation. Methods: All patients with documented acute oesophageal perforation between 1990 and 2009 were enrolled prospectively in an Institutional Review Board-approved database. Results: Of 81 patients who presented during the study period, 52 had upper gastrointestinal endoscopy for diagnosis alone (12 patients; 23 per cent) or as a component of acute management (40 patients; 77 per cent). Use of endoscopy increased from four of 13 patients in the rst 5 years of the study to 20 of 24 patients in the nal 5 years. Endoscopy was used in conjunction with surgery in 28 patients, of whom 21 underwent primary repair, three had resection, and one a diversion; 12 patients in this group had hybrid operations (combination of surgical and endoscopic management). Primary endoscopic treatment was used in 15 patients (29 per cent), most commonly involving stent placement (7). Of those having endoscopy, complication rates improved (from 3 of 4 to 8 of 20 patients), as did mean length of stay (from 218 to 134 days) between the initial and nal 5 years of the study. There were two deaths (4 per cent). Of 21 patients who had both endoscopic assessment and management in the operating room, endoscopy identied additional pathology in ten, leading to a change in management plan in ve patients. Conclusion: Endoscopy is a safe and important component of the management of acute oesophageal perforation. It provides additional information that modies treatment, and its wider use should result in improved outcomes.

Paper accepted 20 December 2010 Published online 29 March 2011 in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.7437

Introduction

Oesophageal perforation is a complex and potentially highly morbid emergency. Mortality rates have historically ranged from 6 to 65 per cent1,2 , although recent publications have documented improved mortality rates in specialized centres3,4 . This improvement may reect new approaches to diagnosis and assessment such as computed tomography, as well as a wider range of management options such as interventional radiology and minimally invasive surgical techniques.
2011 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

Upper gastrointestinal endoscopy has not typically been used in the initial diagnosis and treatment of oesophageal perforation. Some surgeons have considered endoscopy dangerous in the presence of an acute oesophageal injury. Endoscopic techniques now form a recognized component in the management of oesophageal perforation, but in many centres they are conned to use only as part of the initial assessment or in association with the specic intention to initiate endoscopic therapy. The role of endoscopy in diagnosis and treatment of acute oesophageal
British Journal of Surgery 2011; 98: 818824

Endoscopic assessment and treatment in oesophageal perforation

819

perforation has not been clearly dened. This study examined selective, then routine, endoscopy in the management of acute oesophageal perforations over 20 years.

Statistical analysis
Descriptive methods were employed to evaluate clinical characteristics, management and outcome. Group characteristics were compared using Pearsons 2 tests (non-parametric tests) and ANOVA (comparisons of mean values). Statistical signicance was accepted for P < 0050. Statistical analysis was carried out using SPSS version 18 (SPSS, Chicago, Illinois, USA).
Results

Methods

All patients presenting between 1990 and 2009 with oesophageal perforation to a high-volume tertiary referral resident training hospital were enrolled prospectively in an Institutional Review Board-approved database. This database included specic patient demographics, pertinent previous history, and details of presentation, diagnosis and treatment. Patients were included whether they had surgical, non-operative or combined treatment, and included patients initially diagnosed and managed in the authors institution (primary) and those transferred from other hospitals following initial diagnosis and/or treatment (referred). For all patients, diagnosis and management were supervised directly by the surgical team. The present study population included all patients in whom endoscopy was used in the primary diagnostic evaluation or was utilized in conjunction with surgical therapy or to provide primary non-operative treatment. Endoscopic ndings were recorded as a component of the prospective oesophageal perforation database. Specic information recorded included timing and location of the procedure, endoscopic description of the perforation, and secondary pathology. Decision-making associated with endoscopic assessment was made regularly during the endoscopy procedure. As a result, endoscopic interventions were done either by the surgical team or by a gastroenterologist with a member of the surgical team present. All patients with spontaneous or iatrogenic perforations were included. Patients with stulation secondary to oesophageal cancer were excluded from the study.
Table 1

Between April 1990 and May 2009, 81 patients were treated for acute oesophageal perforation at Virginia Mason Medical Center, Seattle. Thirty-three patients (41 per cent) presented at this institution and 48 (59 per cent) were referred from elsewhere. Complications occurred in 21 (44 per cent) of 48 patients who underwent surgery and in ten (30 per cent) of 33 managed without surgery. Overall mean length of hospital stay in non-operative patients was 212 days, compared with 161 days in surgical patients. The endoscopy study group consisted of 52 patients (64 per cent) who had upper gastrointestinal endoscopy as part of their primary assessment or treatment: 22 women and 30 men, with a mean age of 636 (range 2796) years. When the presenting characteristics of patients having endoscopy as a component of their therapy were compared with those of the group without endoscopy, presenting with oesophageal perforation, no signicant differences were noted in age or the proportion of patients presenting for treatment more than 24 h after perforation. The endoscopy group contained a higher proportion of patients with American Society of Anesthesiologists (ASA) grade III or IV, and a longer time period between perforation and initiation of treatment, although these differences were not signicant (Table 1), suggesting that endoscopic interventions were used in patients with more complex presentations.

Characteristics of patients with acute oesophageal perforation who did and did not have endoscopic assessment or treatment
All perforations Endoscopy 52 (64) 636 (2796) 37 (71) 312 (3144) 41 (79) 11 (21) 189 (173) No endoscopy 29 (36) 639 (2289) 18 (62) 262 (2240) 23 (79) 6 (21) 210 (387) P 0942 0402 0690 0564

No. of patients Age (years)* ASA grade IIIIV Time to treatment (h)* Time to diagnosis (h) 24 > 24 Length of stay (days)*

81 (100) 637 (2296) 55 (68) 296 (2240) 64 (79) 17 (21) 197 (187)

0683

Values in parentheses are percentages unless indicated otherwise; *values are mean (range). ASA, American Society of Anesthesiologists. Endoscopy versus no endoscopy (ANOVA), except 2 test.

2011 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery 2011; 98: 818824

820

M. K. Kuppusamy, C. Felisky, R. A. Kozarek, D. Schembre, A. Ross, I. Gan et al.

The most common cause of perforation was iatrogenic (50 per cent), the most common location was the distal oesophagus or oesophagogastric junction (67 per cent) (Table 2), and the size of the mucosal perforation ranged from 1 to 13 cm. In 12 (23 per cent) of the 52 patients, endoscopy was used for diagnosis and assessment only. In 40 (77 per cent), endoscopic intervention included a therapeutic component involving management of the perforation or treatment of secondary pathology. The use of endoscopic techniques over time is shown in Fig. 1, and increased progressively over the study period to include 20 of 24 patients in the last 5 years. Endoscopy was used in 28 of the 48 patients undergoing surgical treatment. Twenty-one patients had primary
Table 2 Presenting characteristics of patients with acute oesophageal perforation who had endoscopy No. of patients (n = 52) Cause of perforation Iatrogenic Therapeutic Diagnostic Barogenic Other Foreign body Food bolus Unknown Location Proximal third Middle third Distal third and gastro-oesophageal junction Size (cm) 1 > 1 to 3 >3 Unknown

26 (50) 24 (46) 2 (4) 19 (37) 7 (13) 2 (4) 3 (6) 2 (4) 9 (17) 8 (15) 35 (67) 12 (23) 15 (29) 10 (19) 15 (29)

repair and drainage, three had a resection, and there was one diversion in a patient with portal hypertension and a 13cm perforation with extensive mediastinal contamination; 12 patients from this group had hybrid procedures involving pleural drainage (open or video-assisted decortication) selectively associated with endoscopic stent placement. In 19 of these 28 patients, endoscopic assessment took place in the operating room immediately before surgery. Endoscopy was used in 24 of the 33 patients who had non-operative management. Fifteen of these 24 patients underwent primary endoscopic therapy, including seven patients undergoing endoscopic stent placement, two undergoing endoscopic therapy in the operating room, two having non-surgical drainage (nasomediastinal drain, 1; interventional radiology drain, 1), two having endoscopic repair with brin glue and endoscopic clips, and nine treated without additional surgical or endoscopic intervention. Nutritional support was augmented with jejunostomy tubes placed during operation in 21 of the 28 surgical patients who underwent endoscopy. Two non-operated patients had endoscopically placed nasojejunal tubes. Six patients had total parenteral nutrition (TPN); in ve patients (3 who initially had TPN) the jejunostomy tube was inserted surgically after primary management. Three surgical and ve non-operative patients did not receive nutritional support but commenced limited oral intake 28 days after treatment. Endoscopic ndings inuenced treatment in 19 (37 per cent) of the 52 patients. Additional ndings were
Table 3 Endoscopic ndings and interventions inuencing surgical treatment of acute oesophageal perforation in patients who had both endoscopic assessment and management in the operating room No. of patients (n = 21) Additional secondary pathology Additional perforation Distal benign stricture Undiagnosed cancer Treatment of secondary pathology Balloon dilatation Stent placement Stent and drain management Nasomediastinal drain Nasojejunal tube placement Previous stent removal 10 2 6 2 6 4 2 6 2 2 2 5 3 1 1

Values in parentheses are percentages.

100 Use of endoscopy (%) 80 60 40 20 0

19901994

19951999 Year

20002004

20052009

Changes in initial planned treatment approach No repair Resection Diversion

Fig. 1

Use of endoscopic techniques over the years of the study

2011 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery 2011; 98: 818824

Endoscopic assessment and treatment in oesophageal perforation

821

Table 4

Evolution of operative, non-operative and endoscopic approaches over the study period
All patients with acute oesophageal perforation Mean LOS (days) Complications 262 210 227 125 197 6 (46) 9 (36) 8 (42) 8 (33) 31 (38)

Operative Mean LOS (days) Complications 26.2 21.8 25.8 13.5 21.2 6 (46) 6 (38) 6 (46) 3 (50) 21 (44)

Non-operative Mean LOS (days) Complications NA 3 (33) 2 (33) 5 (28) 10 (30)

Endoscopy group Mean LOS (days) Complications 218 169 299 134 189 3 (75) 8 (50) 7 (58) 8 (40) 26 (50)

Year 19901994 19951999 20002004 20052009 Entire study period (19902009)

n 13 (27) 16 (33) 13 (27) 6 (13) 48 (100)

n 4 (8) 16 (31) 12 (23) 20 (38) 52 (100)

n 13 (16) 25 (31) 19 (23) 24 (30) 81 (100)

0 (0) NA 9 (27) 196 6 (18) 162 18 (55) 121 33 (100) 161

Values in parentheses are percentages. LOS, length of stay; NA, not applicable.

persistent stricture requiring intraoperative dilatation in six patients, a need for further endoscopic intervention with stent placement, or removal of migrated or malpositioned stents placed earlier at other centres, or intraoperative placement of endoscopic drainage or feeding tubes. Of the 21 patients undergoing endoscopic assessment in the operating room with a tentative plan of management in place, ve had their initial treatment plan changed secondary to ndings at endoscopy (Table 3). Three patients initially thought appropriate for primary repair underwent stenting alone (1 patient), or stenting with surgical drainage of the pleural space and mediastinum when cancer or extensive mucosal necrosis was encountered at endoscopy (2 patients). Another patient was converted to resection when a previously undiagnosed oesophageal malignancy was discovered, and another was converted to diversion following identication of extensive mucosal necrosis and a perforation of over 10 cm in length. Endoscopic interventions in the 28 patients undergoing surgical treatment were employed in two settings. Assessment at the time of presentation was used in 19 patients, or endoscopy was performed in the operating room for initial assessment but also for continued endoscopic guidance during the surgical procedure. Compared with the rest of the surgical group, patients undergoing intraoperative endoscopy had a signicantly higher rate of primary repair (16 of 19 versus 19 of 29; P = 0003) and shorter mean length of stay (208 versus 250 days; P = 0088). The pattern of complications and length of hospital stay in patients having endoscopy as a component of operative and non-operative treatment of oesophageal perforation is shown in Table 4. Of the 52 patients who had endoscopy as a component of assessment or therapy, there were two deaths (4 per cent) at 30 days or at any time in hospital. No recognized
2011 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

complications were directly associated with endoscopic assessment or treatment.

Discussion

Acute oesophageal perforation continues to be a challenging medical emergency. A review of the literature demonstrates wide variations regarding management recommendations. Some high-volume surgical units advocate primary surgical management5 7 , whereas others favour non-surgical therapy3,4,8 . Although non-operative or conservative treatment of acute perforation has been documented for over 50 years9,10 , these non-surgical treatment options historically involved supportive care with nil by mouth, antibiotics, and consideration given to gastric drainage and enteral nutritional support8 . More recently, there has been an increased role for endoscopic and interventional radiology techniques in selected patients2,3,11 17 . This has not only expanded the indications for non-operative therapy, but, as shown in the present experience, has highlighted the evolving role of hybrid treatments where patients are treated either sequentially or, in some cases, synchronously with a combination of surgical and endoscopic interventions. An algorithm for managing patients with acute oesophageal perforation has been developed that highlights the role of endoscopy in initial assessment and denitive treatment (Fig. 2). Although some surgeons may not be familiar with the latest interventional endoscopic interventions, issues regarding the training and availability of endoscopic resources have been resolved in most referral centres. There remains, however, a perception that endoscopy can increase mediastinal contamination, even though this has never been documented and is widely considered irrelevant as long as certain procedural rules are accepted. These include ensuring safe sedation or general anaesthesia to avoid straining or
www.bjs.co.uk British Journal of Surgery 2011; 98: 818824

822

M. K. Kuppusamy, C. Felisky, R. A. Kozarek, D. Schembre, A. Ross, I. Gan et al.

Acute oesophageal perforation

Clinically stable

Unstable/septic

Water soluble/barium contrast study (observed by managing team)

If feasible

Non-intubated

Intubated

Followed immediately by CT

CT followed by OGD by managing team

Contained perforation

Contained perforation or limited free perforation

Extensive contained perforation or free perforation Decision favours endoscopic therapy or patient too unstable for surgery

Decision favours observation

Decision favours endoscopic therapy

Decision favours surgical therapy

Nil by mouth Antibiotics/ antifungals NG tube nutritional support

OGD in OR before positioning Leave scope in place during operative procedure

If primary repair, review options for additional endoscopic therapy

Endoscopic clips/stents Transoesophageal drainage Enteric feeding tubes additional surgical or interventional radiology drains
Fig. 2

Algorithm for the management of acute oesophageal perforation. CT, computed tomography; OGD, oesophagogastroduodenoscopy; NG, nasogastric; OR, operating room

retching during the endoscopic procedure. The procedure should be carried out in the intensive care unit, designated endoscopy suite or operating room, where a decision regarding intubation and general anaesthesia can readily be facilitated and intrapleural drains inserted if required. A number of general points have emerged during the evolution of the present series. If endoscopy is done in conjunction with surgical repair, initial assessment is best carried out immediately after initiation of general anaesthesia, but before positioning. Air insufation should be minimized, although appropriate visibility is critical to support decisions regarding positioning, approach and ultimate management. Endoscopic assessment of mucosal viability is probably more important than the historical timeframe issues of initial treatment within 24 h.
2011 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

Once the initial assessment is complete, the endoscope is left in place in the stomach to guide intraoperative therapy. In addition to the recognition of secondary pathologies that can modify management strategies, perioperative endoscopy conrms the location and extent of the mucosal defect, and suitability for stents, endoscopic clips or transoesophageal drains as a component of denitive or hybrid approaches to repair. The indwelling endoscope can guide the surgeon specically to the mucosal perforation, an aspect that can be particularly important when managing delayed perforations (Fig. 3a). Endoscopic visualization can also guide repair sutures in certain circumstances (Fig. 3b), and can be used to insufate the oesophagus in order to test primary repairs or facilitate the safe placement of enteric drains or feeding tubes.
www.bjs.co.uk British Journal of Surgery 2011; 98: 818824

Endoscopic assessment and treatment in oesophageal perforation

823

It seems likely that endoscopy will play an increasing role in the assessment and primary management of oesophageal perforation. Hybrid procedures should become more common as surgical teams accept endoscopy as a critical part of diagnosis and treatment, and become increasingly familiar with interventional endoscopic techniques.
Acknowledgements

The authors would like to acknowledge the help of Dr Philip Carrott and Bonnie A. Marston in the preparation of the manuscript. The authors declare no conict of interest.
References

Assessment

Surgery

Fig. 3

a Intraoperative endoscopy for the assessment of mucosal viability and precise localization of the mucosal defect. This intervention is particularly important in patients with a delayed presentation, when intramural and mediastinal contamination can obscure the anatomy. The illustration demonstrates the potential for intramural dissection of blood and contamination to be separate from the site of mucosal injury. b Intraoperative endoscopy to guide primary surgery (Illustration by Joanne Clifford)

When the management of all perforations over the 19 years of the study are considered, patients who had endoscopy as a component of diagnosis or treatment had a higher ASA grade and were more likely to have delayed management, indicating that endoscopic intervention was not reserved for patients with more straightforward presentations. The importance placed on preoperative endoscopy is highlighted by the fact that it was used in four of 13 patients in the rst 5 years, compared with 20 of 24 in the last 5 years. Endoscopy in conjunction with operation merits particular comment. Nineteen of the 28 patients in this group had synchronous endoscopic assessment and management. Endoscopy was specically responsible for identifying additional important diagnoses in ten patients, including additional perforations in two, undiagnosed strictures in six and cancer in two patients. This resulted in a direct modication of treatment plans in six of these patients (balloon dilatation, 4; stent placement, 2). Additional interventions were also used in six other patients (removal of pre-existing stents, 2; insertion of nasomediastinal drains, 2; and nasojejunal feeding tubes, 2). Notably, the planned operative therapy was changed signicantly in ve surgically managed patients.
2011 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

1 Kim-Deobald J, Kozarek RA. Esophageal perforation: an 8-year review of a multispecialty clinics experience. Am J Gastroenterol 1992; 87: 11121119. 2 Huber-Lang M, Henne-Bruns D, Schmitz B, Wuerl P. Esophageal perforation: principles of diagnosis and surgical management. Surg Today 2006; 36: 332340. 3 Port JL, Kent MS, Korst RJ, Bacchetta M, Altorki NK. Thoracic esophageal perforations: a decade of experience. Ann Thorac Surg 2003; 75: 10711074. 4 Vogel SB, Rout WR, Martin TD, Abbitt PL. Esophageal perforation in adults: aggressive, conservative treatment lowers morbidity and mortality. Ann Surg 2005; 241: 10161021. 5 Jones WG II, Ginsberg RJ. Esophageal perforation: a continuing challenge. Ann Thorac Surg 1992; 53: 534543. 6 Brinster CJ, Singhal S, Lee L, Marshall MB, Kaiser LR, Kucharczuk JC. Evolving options in the management of esophageal perforation. Ann Thorac Surg 2004; 77: 14751483. 7 Richardson JD. Management of esophageal perforations: the value of aggressive surgical treatment. Am J Surg 2005; 190: 161165. 8 Siersema PD. Treatment of esophageal perforations and anastomotic leaks: the endoscopist is stepping into the arena. Gastrointest Endosc 2005; 61: 897900. 9 Michel L, Grillo HC, Malt RA. Operative and nonoperative management of esophageal perforations. Ann Surg 1981; 194: 5763. 10 Kiss J. Surgical treatment of oesophageal perforation. Br J Surg 2008; 95: 805806. 11 Grifn SM, Lamb PJ, Shenne J, Richardson DL, Karat D, Hayes N. Spontaneous rupture of the oesophagus. Br J Surg 2008; 95: 11151120. 12 Eroglu A, Turkyilmaz A, Aydin Y, Yekeler E, Karaoglanoglu N. Current management of esophageal perforation: 20 years experience. Dis Esophagus 2009; 22: 374380.

www.bjs.co.uk

British Journal of Surgery 2011; 98: 818824

824

M. K. Kuppusamy, C. Felisky, R. A. Kozarek, D. Schembre, A. Ross, I. Gan et al.

13 Vallbohmer D, Holscher AH, Holscher M, Bludau M, Gutschow C, Stippel D et al. Options in the management of esophageal perforation: analysis over a 12-year period. Dis Esophagus 2010; 23: 185190. 14 Abbas G, Schuchert MJ, Pettiford BL, Pennathur A, Landreneau J, Landreneau J et al. Contemporaneous management of esophageal perforation. Surgery 2009; 146: 749755. 15 Freeman RK, Van Woerkom JM, Ascioti AJ. Esophageal stent placement for the treatment of iatrogenic intrathoracic

esophageal perforation. Ann Thorac Surg 2007; 83: 20032007. 16 Fischer A, Thomusch O, Benz S, von Dobschuetz E, Baier P, Hopt UT. Nonoperative treatment of 15 benign esophageal perforations with self-expandable covered metal stents. Ann Thorac Surg 2006; 81: 467472. 17 Karbowski M, Schembre D, Kozarek R, Ayub K, Low D. Polyex self-expanding, removable plastic stents: assessment of treatment efcacy and safety in a variety of benign and malignant conditions of the esophagus. Surg Endosc 2008; 22: 13261333.

Snapshot Quiz 11/03

Question: What is this lesion and how should it be treated?

The answer to the above question is found at the end of the Your Views section in this issue of BJS.
Teo I, Wild JRL, Goodfellow PB: Chestereld Royal Hospital, Calow, Chestereld, Derbyshire S44 5BL, UK (e-mail: isabelteo@hotmail.com)

Snapshots in Surgery: to view submission guidelines, submit your snapshot and view the archive, please visit www.bjs.co.uk

2011 British Journal of Surgery Society Ltd Published by John Wiley & Sons Ltd

www.bjs.co.uk

British Journal of Surgery 2011; 98: 818824