Volume 133, Issue 2 , Pages 333-338, February 2007
Postoperative esophageal leak management with the Polyflex esophageal stent
Article Outline
Objective
Leak after esophageal anastomosis or perforation repair prolongs hospitalization, prevents oral hydration and nutrition, and can produce localized infection or sepsis. This investigation reviews our experience treating postoperative esophageal leaks with the Polyflex esophageal stent (Boston Scientific, Natick, Mass).
Methods
Over a 30-month period, patients with a postoperative esophageal leak were treated with the Polyflex stent for leak occlusion. Leak occlusion was confirmed by means of esophagraphy. Patients were followed until their stent was removed and their esophageal leak had resolved.
Results
Twenty-one patients had 27 stents placed for leak occlusion after esophagectomy (n = 5), esophageal perforation (n = 5), surgical (n = 4) or endoscopic (n = 2) antireflux procedure, and esophageal diverticulectomy (n = 3) or myotomy (n = 2). The mean interval between surgical intervention and stent placement was 12 ± 8 days (range, 3–31 days). Occlusion of the leak occurred in 20 patients. One patient experienced a dehiscence of the surgical esophageal perforation repair requiring esophageal diversion. Stent migration requiring repositioning (n = 3) or replacement (n = 4) occurred in 5 (24%) patients. Twenty (95%) stents were removed without residual leak (mean, 51 ± 43 days; range, 15–175 days). One patient had a stricture after stent removal that required endoscopic dilatation. One patient in this series died.
Conclusions
The Polyflex esophageal stent is an effective method for occluding a postoperative esophageal leak. It rapidly eliminates contamination of the mediastinum, pleura, and peritoneum; allows oral hydration and nutrition; and is easily removable. These stents also offer an appealing alternative to traditional esophageal diversion and subsequent reconstruction in patients with a persistent esophageal leak.
CTSNet classification: 8
Leak after esophageal anastomosis or perforation repair can be a challenge for the surgeon. Traditional therapy for a symptomatic postoperative esophageal fistula has been operative repair or, if unsuccessful, esophageal diversion. Such treatment prolongs hospitalization and significantly delays oral hydration and nutrition. In the case of subsequent unsuccessful repairs, esophageal diversion commits the patient to further surgical intervention to re-establish foregut continuity.
Endoluminal esophageal stents have been available for many years and have been used to palliate unresectable esophageal malignancies with dysphagia. Unfortunately, the characteristics of the material of which stents were made caused them to be difficult to insert and remove and resulted in significant complications, such as stent migration, fistula formation or enlargement, and bleeding. However, since esophageal stents made of materials that make them easier to place and remove, are significantly more flexible, and exert less radial force and remain occlusive have become available, recent reports of their use to treat symptomatic esophageal anastomotic leak after cancer resection have been published.1, 2 This investigation reviews our experience treating recalcitrant postoperative esophageal fistulae endoscopically using an occlusive, silicon-coated esophageal stent (Polyflex; Boston Scientific, Natick, Mass).
Materials and Methods
Over a 30-month period, patients who experienced a postoperative esophageal leak at either of 2 tertiary-care hospitals were offered participation in an institutional review board–approved protocol using an endoluminal esophageal stent to treat their fistula, a use of the Polyflex stent not currently recognized by the Food and Drug Administration. Patients with a documented esophageal leak that had recurred after at least 1 operative repair were eligible for participation in this investigation. For the purposes of this investigation, patients with an esophageal perforation or esophageal malignancies not previously treated with surgical intervention were excluded. Also excluded from participation were patients with an anastomotic disruption or ischemic esophageal conduit that would necessitate esophageal diversion. The presence of an esophageal fistula was documented and localized by means of gastrograffin esophagraphy, barium esophagraphy, or both before any treatment. Additionally, all patients being considered for stent placement underwent computer-aided tomographic imaging of the neck, chest, and abdomen.
All esophageal stents were placed in the operating room by a thoracic surgeon using general endotracheal anesthesia and fluoroscopy after flexible esophagoscopy. Adequate drainage of infected areas was also simultaneously achieved. Fistula occlusion was confirmed by means of contrast esophagraphy a minimum of 48 hours after stent placement or when the patient was able to participate in the examination. In the absence of a continued leak, patients who were able were begun on a “soft mechanical” diet without bread or meat and advanced as tolerated to a “postgrastrectomy” diet.
It was the intention to remove all patients’ esophageal stents after a sufficient amount of time to allow permanent fistula closure. This was individualized based on the fistula cause, anatomic location, nutritional status of the patient, and resolution of all associated infectious or septic parameters, including the resolution of any associated organ failure, but was not less than 14 days after initial stent placement. Stent removal again was carried out in the operating room after achievement of general anesthesia. Flexible esophagoscopy was performed before and after stent removal as well. An esophagraphy was again performed 24 hours after stent removal to exclude recurrent leak. Patients were followed until their esophageal fistula had resolved, their stent had been removed, and they were tolerating oral nutrition. All patients were assessed at least 1 month after stent removal for dysphagia. Continuous data for this investigation are expressed as the mean ± standard deviation of the mean, except where otherwise indicated.
Results
During the 30-month study period, 29 patients were evaluated by the authors at either of the 2 study institutions for a postoperative esophageal fistula. Twenty-two patients met the investigation’s inclusion criteria; 1 of these patients refused further intervention and was not included in this review. Thus 21 patients were available to participate. Follow-up was complete in all of these 21 patients for the study period. Their mean age was 58 ± 20 years (range, 14–93 years). The causes of their esophageal fistulae were esophagectomy (n = 5), esophageal perforation repair (n = 5), surgical antireflux procedure (n = 4), esophageal diverticulectomy (n = 3), endoscopic antireflux procedure (n = 2), and esophageal myotomy (n = 2, Table 1). Five patients were seen in consultation as inpatients after their original operation, and 16 patients were transferred from other hospitals before their evaluation. The mean number of attempted repairs before stent placement was 1.4 ± 0.8 (range, 1–4; median, 1). The mean number of days from initial surgical repair to stent placement was 12 ± 8 days (range, 3–31 days). Twelve (57%) patients underwent additional procedures at the time of stent placement (Table 2), the most common of which was for enteral feeding access.
Table 1. Patient demographic
| Age (y) | Initial procedure | Attempted operative repairs | Fistula location |
|---|---|---|---|
| 64 | Esophagectomy | 1 | Pleura |
| 59 | Esophagectomy⁎ | 2 | Trachea |
| 51 | Esophagectomy⁎ | 1 | Bronchus |
| 67 | Esophagectomy⁎ | 1 | Pleura |
| 54 | Esophagectomy⁎ | 1 | Bronchus |
| 44 | Perforation repair | 3 | Cutaneous (neck) |
| 93 | Perforation repair | 1 | Peritoneum |
| 69 | Perforation repair | 1 | Pleura and peritoneum |
| 82 | Perforation repair | 1 | Bronchus |
| 84 | Perforation repair | 2 | Mediastinum |
| 34 | Surgical antireflux | 1 | Peritoneum |
| 48 | Surgical antireflux | 1 | Mediastinum |
| 39 | Surgical antireflux | 1 | Peritoneum |
| 49 | Surgical antireflux | 2 | Pleura and peritoneum |
| 73 | Diverticulectomy | 1 | Pleura |
| 66 | Diverticulectomy | 1 | Mediastinum |
| 37 | Diverticulectomy | 2 | Mediastinum |
| 72 | Endoscopic antireflux | 1 | Mediastinum |
| 78 | Endoscopic antireflux | 1 | Peritoneum |
| 33 | Esophageal myotomy | 1 | Peritoneum |
| 14 | Esophageal myotomy | 4 | Peritoneum and pleura |
⁎Received preoperative chemotherapy and external beam radiation therapy. |
Table 2. Surgical procedures performed in addition to esophageal stent placement
| Associated surgical procedures | |
|---|---|
| Tube jejunostomy | 10 |
| Tube thoracostomy | 5 |
| Percutaneous endoscopic gastroscopy | 4 |
| Thoracoscopic decortication | 4 |
| Tracheostomy | 3 |
| Celiotomy | 2 |
| Neck exploration | 1 |
| Total | 29 |
Mean days from stent placement until repeat esophagraphy was 6 ± 7 days (range, 2–29 days; median, 4 days). The most common reason for delay of esophagraphy beyond 48 hours was continued mechanical ventilation. Occlusion of the esophageal fistula occurred in 20 (95%) patients, as demonstrated by the initial esophagram. All 20 patients were able to be started on an oral intake regimen, as outlined above. Polyflex stents in 20 (95%) patients were removed without residual esophageal leak or fistula at a mean of 51 ± 43 days (range, 15–175 days). None of these patients required further operations for their esophageal fistula. Hospital length of stay varied widely because of the cause of the esophageal fistula and associated complications. The mean was 12 ± 12 days (range, 4–44 days; median, 7 days).
All stent placements occurred without intraoperative complications. One patient experienced the dehiscence of the esophageal perforation repair 2 days after stent placement and required operative repair. Other associated morbidities included respiratory failure (n = 5), pneumonia (n = 2), and deep venous thrombosis (n = 1), and 1 patient required celiotomy for an enterocutaneous fistula thought to be related to the original abdominal operation (Table 3). One (5%) patient had an esophageal stricture after stent removal that required endoscopic dilatation. One (5%) patient died from the rupture of an infected thoracic aortic graft after the reoperative repair of an aortic dissection complicated by esophageal perforation. This patient’s esophageal injury was recognized and repaired by means of primary closure of the esophagus and pedicled muscle interposition at the time of the aortic dissection repair. When the leak recurred, endoluminal esophageal stent placement was performed. This patient’s stent had been removed before death.
Table 3. Morbidities occurring after stent placement
| Morbidities | |
|---|---|
| Stent migration | 5 (24%) |
| Respiratory failure | 3(14%) |
| Pneumonia | 2(10%) |
| Deep venous thrombosis | 1(5%) |
| Enterocutaneous fistula | 1(5%) |
Discussion
The use of an endoluminal esophageal stent is not a new concept. Esophageal intubation has been used since the 19th century, when Symonds3 described his experience with prostheses made of ivory and silver. In more recent times, Mousseau and Celestin, and Atkinson and Ferguson have all developed devices for esophageal intubation.4, 5, 6 Difficulty with insertion, migration, and extraction, however, limited the use of these prostheses to the palliation of unresectable malignant conditions of the esophagus, with rare exceptions.7
Taking advantage of the technology used to make endovascular stents, self-expanding metallic esophageal stents became available in the 1990s. These stents could be inserted with flexible esophagoscopy, required significantly less esophageal dilatation, had a lower rate of migration, and provided excellent palliation for malignant esophageal strictures.8 Subsequently, occlusive self-expanding esophageal stents were used to treat patients with malignant tracheoesophageal fistulae, as well as esophageal obstruction.9 The silicone covering decreased tumor ingrowth while forming an occlusive seal in the case of a fistula. However, there continued to be reluctance to place these prostheses in the esophagus of a patient for conditions other than palliative therapy for a malignancy because of the potential esophageal damage associated with extraction.
The recent ability to produce a plastic prosthesis coated with silicone has resulted in an esophageal stent that has a distinct advantage over previous versions. Ease of insertion, the requirement for minimal esophageal dilation, and the formation of an occlusive seal within the lumen of the esophagus are all similar to the self-expanding metallic stents. The distinct advantage of these nonmetallic endoprostheses is their unique ability to be extracted without damaging the esophagus. This has led some investigators to implant these stents in select patients as a temporary measure to treat an esophageal leak or fistula after esophagectomy.10
Patients who experience a continued leak or have an esophageal fistula after initial repair of an esophageal perforation or anastomotic dehiscence will most often experience a prolonged hospital course, realize a significant delay in their ability to take oral nutrition and hydration, and have the potential to experience localized infectious complications, as well as sepsis. The traditional treatment for these patients has been a reoperative attempt at repair or esophageal diversion with or without esophagectomy. On the basis of our anecdotal experiences using an occlusive, silicone-coated esophageal stent (Polyflex), we hypothesized that temporary esophageal stent placement could effectively treat some postoperative esophageal fistulae without the need for further esophageal operations.
This investigation reviews our experiences with a large group of patients treated in this manner. We found that endoluminal esophageal stent placement was easily accomplished. This resulted in the rapid closure of postoperative esophageal fistulae, eliminating this source of infection/sepsis in all but 1 patient in this series. Although some patients could not attempt oral nutrition because of comorbid processes, patients who could otherwise tolerate an oral diet began one significantly sooner than would be expected after operative fistula repair. Finally, although difficult to quantify, it would appear that the 20 patients who were successfully treated with esophageal stent placement should have benefited from a reduction in the potential morbidity associated with another major surgical procedure. Similarly, a small percentage of these 20 patients could have required esophageal diversion with subsequent reconstruction, neither of which was realized because of endoluminal therapy.
These results are in keeping with 2 smaller series of patients with a postoperative esophageal anastomotic leak who received esophageal stent placement instead of operative repair.11, 12 However, the current series is unique in both its size and the treatment algorithm used. Both previously reported series each contained significantly fewer patients than the series presented in this investigation. Furthermore, neither of the cited series treated a postoperative esophageal fistula definitively with surgical intervention, as was our standard, followed by endoluminal esophageal stent placement only if the fistula recurred. This strategy reflects the likely preference of the majority of thoracic surgeons and identifies a viable alternative to further surgical intervention.
The migration of these stents does minimally detract from their use. In this review 7 stents migrated in 5 patients, requiring repositioning or replacement. This compares favorably with other series in which esophageal stents have been used to palliate malignant strictures.13 However, we have recognized that this can be minimized by slightly oversizing the diameter and significantly oversizing the length of the stent when possible. Migration in our series was also more common in the proximal cervical esophagus and at the gastroesophageal junction, as would be expected because of the decreased surface area of the esophagus available to seat the stent.
Esophageal stricture formation after stent removal also occurred in 1 (5%) patient in this investigation. This has been reported previously and appears to occur most often if the leak being treated is at the site of an esophagogastrostomy. In the series by Langer and colleagues,2 as well as in our experience, such strictures are easily treated without further surgical intervention and should not detract significantly from the use of these stents.
Similarly important to the success of treating patients with a postoperative esophageal fistula by using an esophageal stent is to achieve the same goals one would entertain if operative repair was performed. In addition to fistula occlusion, these include eliminating sources of infection/sepsis, establishing reliable access for enteral nutrition, and minimizing postoperative morbidity. In this series patients required an average of 1.4 simultaneous procedures to accomplish these goals. We believe these associated procedures were integral to whatever success we might have achieved. In our opinion this global approach to these patients can only be provided with the involvement of and preferably under the direction of a thoracic surgeon.
Although it would appear that endoluminal esophageal stent placement for postoperative esophageal fistulae could have some advantages over further attempts at operative repair, this investigation has some weaknesses. Although representing the largest number of postoperative esophageal fistulae treated in this manner in the current literature, a patient population of 21 remains a small number, even for a relatively uncommon condition. Furthermore, no attempt was made to compare endoluminal therapy with reoperative therapy in either a prospective or retrospective fashion. It was believed that such a comparison would be of minimal benefit because of the significant variation in patients’ comorbidities at the time the fistula was recognized, as well as the varied nature of their underlying esophageal diseases.
In conclusion, this series demonstrates the effectiveness of an occlusive esophageal stent in treating patients with recalcitrant postoperative esophageal fistulae. Endoluminal stent placement provides rapid fistula closure, eliminating soilage of the mediastinum, pleura and, peritoneum; allows patients to begin oral intake within 48 hours of stent placement; and eliminates the need for further operations in the vast majority of patients. Stent placement also offers an appealing alternative to reoperative repair and especially esophageal diversion and subsequent reconstruction. Proper patient selection and stent sizing should minimize the principle weakness of this treatment strategy: stent migration. The initial success demonstrated by this review might hold promise for treating selected postoperative esophageal fistulae, as well as iatrogenic esophageal perforations, with endoluminal techniques.
References
- . Management and results of proximal anastomotic leaks in a series of 1114 total gastrectomies for gastric carcinoma. Eur J Surg Oncol. 2000;26:168–171
- Management of postoperative esophageal leaks with a Polyflex self-expanding covered plastic stent. Ann Thorac Surg. 2005;79:398–404
- . The treatment of malignant stricture of the oesophagus by tubage or permanent catheterism. BMJ. 1887;1:870–881
- . Place de l’intubation a demeure dans le traitement palliative du cancer de l’oesophage. Arch Fr Mal Appl Dig. 1956;45:208–215
- . Permanent intubation in inoperable cancer of the esophagus and cardia: a new tube. Ann R Coll Surg. 1959;25:165–172
- . Fiberoptic endoscopic palliative intubation of inoperable oesophagogastric neoplasms. BMJ. 1977;1:266–273
- . Palliative intubation in the management of esophageal carcinoma. Ann Thorac Surg. 1974;18:417–425
- . Treatment of esophageal obstruction with covered, self-expanding esophageal wall stents. Ann Thorac Surg. 1996;62:963–969
- . Covered, expandable esophageal metallic stent tubes: experiences in 119 patients. Radiology. 1994;193:689–693
- Use of self-expandable plastic stents for the treatment of esophageal perforations and symptomatic anastomotic leaks. Endoscopy. 2004;36:695–699
- . Treatment of thoracic anastomotic leaks after esophagectomy with self-expanding plastic stents. Ann Surg. 2004;240:8801–8807
- Endoscopic treatment of thoracic esophageal anastomotic leaks by using silicone-covered, self-expanding polyester stents. Gastrointest Endosc. 2005;61:897–900
- . Use of the Polyflex stent in the palliative therapy of esophageal carcinoma: results in 14 cases and review of the literature. Surg Endosc. 2001;15:14444–14447
Dr Freeman
PII: S0022-5223(06)01851-4
doi:10.1016/j.jtcvs.2006.10.008
© 2007 The American Association for Thoracic Surgery. Published by Elsevier Inc. All rights reserved.
Volume 133, Issue 2 , Pages 333-338, February 2007
