Management of anastomotic leaks after Roux-en-Y bypass using self-expanding polyester stents

Surgery for Obesity and Related Diseases 4 (2008) 594 – 600

Original article

Management of anastomotic leaks after Roux-en-Y bypass using self-expanding polyester stents Christopher A. Edwards, M.D., Phiet T. Bui, M.D., J. Andrés Astudillo, M.D., Roger A. de la Torre, M.D., F.A.C.S., Brent W. Miedema, M.D., F.A.C.S.*, Archana Ramaswamy, M.D., F.A.C.S., Nicole M. Fearing, M.D., F.A.C.S., Bruce J. Ramshaw, M.D., F.A.C.S., Klaus Thaler, M.D., F.A.C.S., J. Stephen Scott, M.D., F.A.C.S. Department of Surgery, University of Missouri–Columbia, Columbia, Missouri Received June 9, 2007; revised February 27, 2008; accepted May 1, 2008

Abstract

Background: To analyze the outcomes of a series of endoscopically placed polyester self-expanding polyflex stents (SEPSs) for the management of anastomotic leaks after Roux-en-Y bypass. Anastomotic leaks after gastric bypass cause significant morbidity and mortality. Covered polyester SEPSs might have a role in the treatment of these leaks. Methods: A retrospective chart review was performed from January 2006 to November 2006 that included all acute and chronic leaks treated with SEPSs. Results: A total of 6 patients were treated with stents, with a mean procedure time of 22 minutes. Of these 6 patients, 5 had acute postoperative leaks and 1 had a chronic fistula. Five patients started oral intake 1– 6 days after their procedure. All acute leaks had complete healing at a median of 44 days. The patient with a chronic gastrocutaneous fistula required revisional surgery for fistula closure. In addition, 5 patients had stent migration, and 3 required stent replacement. Conclusion: An endoscopically placed SEPS provides a less-invasive alternative to treat acute anastomotic leaks after Roux-en-Y bypass while simultaneously allowing oral intake. The results of this case series have demonstrated this treatment to be safe and effective. (Surg Obes Relat Dis 2008;4:594 – 600.) © 2008 American Society for Metabolic and Bariatric Surgery. All rights reserved.

Keywords:

Gastric bypass; Anastomotic leak; Endoscopy; Stent

The incidence of anastomotic leaks after Roux-en-Y gastric bypass (RYGB) procedures ranges from 0.3% to 8.3% and has been similar in both open and laparoscopic series [1]. Anastomotic leaks at the gastrojejunostomy (GJ) or esophagojejunostomy (EJ) after Roux-en-Y bypass can be difficult to treat and can result in significant morbidity and mortality. Therefore, it is essential to im-

*Reprint requests: Brent W. Miedema, M.D., F.A.C.S., University of Missouri, One Hospital Drive, MC422 McHaney Hall, DC075.00, Columbia, MO 65212. E-mail: [email protected]

prove the methods of treating anastomotic leaks after RYGB. The conventional management of anastomotic leaks after RYGB has historically been surgical [2]. The tenets of surgery include washout of the peritoneal cavity, appropriate debridement and repair of the anastomotic leak, followed by drainage of this region. Drainage modalities range from percutaneous radiographically guided drainage, open surgical drainage, or intraluminal T-tube placement at the leak site. Laparoscopic or open approaches can be performed according to the surgeon’s expertise and the clinical situation. In a subset of hemodynamically stable and asymptomatic patients, nonoperative expectant management can be used. Within these scenarios, most

1550-7289/08/$ – see front matter © 2008 American Society for Metabolic and Bariatric Surgery. All rights reserved. doi:10.1016/j.soard.2008.05.009

C. A. Edwards et al. / Surgery for Obesity and Related Diseases 4 (2008) 594 – 600

patients require restriction of oral intake across the anastomosis, intravenous antibiotics, and specialized nutritional support. The use of endoscopically placed, covered, polyester, selfexpanding polyflex stents (SEPSs) is a novel approach for the treatment of anastomotic leaks after RYGB. This procedure provides a minimally invasive alternative within a broad spectrum of modalities. These stents exclude the leak from intestinal contents, allowing oral nutritional intake while the leak heals, and thus avoids the morbidity from specialized nutrition. In our institution, we have successfully managed selected GJ and EJ anastomotic leaks after Roux-en-Y bypass procedures with silicone-covered SEPSs. The purpose of this study was to present the outcomes of our series of patients. Methods A retrospective chart review was performed at a single tertiary bariatric center of consecutive patients with a GJ or an EJ anastomotic leak after laparoscopic or open RYGB treated from January 2006 to November 2006 with a covered SEPS. Pertinent demographic data were collected retrospectively. Patients with acute postoperative leaks and chronic fistula from leaks were included in the study. The

595

GJ and EJ anastomotic leaks in all the patients were identified using both clinical and radiographic criteria. Algorithms for treating anastomotic leaks included drainage for all acute leaks and local wound care for chronic fistulas. In each case, the SEPS (Polyflex, Boston Scientific, Natick, MA) was placed using endoscopic and fluoroscopic guidance under general anesthesia in the operating room. Upper endoscopy was used to identify the intraluminal location of the GJ or EJ leak, and metallic markers were externally placed on the patient’s body to mark the location of the leak under fluoroscopy. The SEPS was then deployed under fluoroscopic guidance using the metallic markers as the primary landmarks. An upper gastrointestinal (UGI) contrast study was performed after the procedure during the same hospital stay. Patients without an identifiable leak on the UGI contrast study were then allowed to start a standard clear liquid bariatric diet and advance according to the protocol used for bariatric patients in our institution. The primary endpoints after stent placement included radiographic evidence of leak after stent placement, symptom improvement, and leak closure. The secondary endpoints included procedure time, stent migration, and stentrelated morbidity. All patients were followed up clinically

Table 1 Initial operation, clinical presentation, and leak management Pt. No.

Initial operation before leak

Clinical presentation of leak

Management

1

Open revision of RYGB (takedown of gastrogastric and gastrocutaneous fistula; repair of ventral hernia)

Endoscopy on POD 6, with placement of SEPS; chronic fistula developed and was resolved with stent treatment

2

Open revision of vertical banded gastroplasty to RYGB (splenectomy [splenic bleeding]; jejunostomy tube placement; positive air leak test) Open esophagojejunostomy (RYGB initially; intraoperative endoscopy demonstrated iatrogenic partial gastric inlet obstruction)

Symptoms: delirium, leukocytosis, normal vital signs Workup: abdominal CT showing free contrast; UGI study showing contrast leak Symptoms: none Workup: UGI study showing contrast leak Symptoms: left shoulder pain, normal vital signs, leukocytosis; no abdominal pain Workup: abdominal CT showing free air/fluid; UGI study positive for leak Symptoms: tachycardia, tachypnea, leukocytosis Workup: abdominal CT showing free air and contrast leak; UGI study showing contrast leak Symptoms: respiratory distress requiring intubation Workup: UGI study showing contrast leak on POD 20

Exploratory laparotomy on POD 7, with abdominal washout and intra-abdominal drains placed Endoscopy on POD 10, with placement of SEPS

3

4

Primary laparoscopic RYGB with negative air leak test

5

Laparoscopic revision RYGB (takedown gastrogastric fistula; remnant gastrectomy)

6

Open revision RYGB 11 mo before leak

Symptoms: left shoulder pain; chronic cutaneous fistula drainage Workup: abdominal CT showing subdiaphragmatic abscess

Endoscopy on POD 4, with placement of SEPS

Diagnostic laparoscopy on POD 4, with laparoscopic washout and intra-abdominal drains placed Postoperative endoscopy on POD 28, with placement of SEPS Exploratory laparotomy on POD 6, with abdominal washout, closure of duodenal stump, and intra-abdominal drains placed Postoperative endoscopy on POD 27, with placement of SEPS Endoscopy, with placement of SEPS; multiple replacements required Open revision of RYGB to esophagojejunostomy 7 mo after stent placement

Pt. No. ⫽ patient number; RYGB ⫽ Roux-en-Y gastric bypass; CT ⫽ computed tomography; UGI ⫽ upper gastrointestinal; POD ⫽ postoperative day; SEPS ⫽ self-expanding polyflex stent.

596

C. A. Edwards et al. / Surgery for Obesity and Related Diseases 4 (2008) 594 – 600

for recurrent symptoms. Before and after stent removal, radiographic imaging with an UGI contrast study was performed to confirm the resolution of the anastomotic leak. Standard descriptive data analysis was performed.

Results From January 2006 to November 2006, 6 patients presented with a GJ or an EJ anastomotic leak after Roux-en-Y bypass. The patients were all women, with mean patient age of 50 years (range 32– 60). Five patients had developed acute leaks within 20 postoperative days after their bypass. Table 1 lists the type of operation performed before the anastomotic leak developed, the clinical presentation, and the definitive treatment. Patient 1 developed a chronic gastrocutaneous fistula after stent placement for her acute leak that was also managed by endoscopic SEPS placement. Patient 6 presented with a chronic gastrocutaneous fistula. Three patients underwent surgical abdominal washouts because of intra-abdominal fluid collections, with placement of a large, soft, closed-suction drain placed next to the anastomotic leak. Fig. 1 shows a computed tomography scan of the abdomen of 1 patient with the findings of free air and fluid surrounding a perforated GJ anastomosis. In 3 patients, endoscopic stent placement was the only treatment needed for the leak; all of these patients were hemodynamically stable. Delayed endoscopic stenting was performed after surgical control of the intra-abdominal infection in 3 patients because surgical repair of the inflamed anastomotic leaks was thought to be unlikely to succeed. Fig. 2A–C show fluoroscopic images of the same patient with a GJ anastomotic leak, with a stent placed across the leak, and with a healed anastomosis after stent removal. Table 2 summarizes the stent outcome data. The mean operative time for stent placement was 23 minutes. The SEPSs were in place for a mean of 35 days in the acute leak group. In the patients with a chronic leak, including the patient with an acute leak who developed a chronic fistula, the stent remained in place for a mean of 305 days. All patients had symptomatic improvement after stent placement. Of the 6 patients, 5 resumed oral feeding 1– 6 days (mean 4) after stent placement. Patient 5 had poor motility of the Roux limb before revisional surgery and did not tolerate oral feeding despite normal fluoroscopy findings. She required enteral feeding but was able to covert to oral feeding 2 months after stent placement. No major stent-related morbidity or mortality developed. Stent migration occurred in 5 (83%) of the 6 patients. The stent had to be replaced in 3 of these patients. Patient 1 required 2 stent replacements for her acute leak and 2 for her chronic fistula. Stent migration was usually only a few centimeters but was enough to expose the anastomotic leak. These stents were removed and usually replaced with new stents endoscopically. Patient 4 had 2 separate stents mi-

Fig. 1. Abdominal and pelvic computed tomography scan demonstrating free air and fluid (white arrow) around gastrojejunal anastomotic leak.

grate distally in the small bowel. The second stent was placed to cover the anastomotic leak, but it subsequently migrated as well. Both stents were monitored with serial radiography until passed per rectum without further complications or the necessity for endoscopic retrieval. In the first 4 patients, the initial stents placed had a 16 mm diameter and 9 or 12 cm length. All these stents migrated and were subsequently replaced with stents 18 mm in diameter and 15 cm long. One of the 15-cm stents migrated before closure of the anastomotic leak and required endoscopic retrieval.

C. A. Edwards et al. / Surgery for Obesity and Related Diseases 4 (2008) 594 – 600

597

Fig. 2. (A) Radiograph of anastomotic leak. White arrow shows contrast extravasation. (B) Radiograph of same patient showing stent in place bypassing anastomotic leak. (C) Radiograph of healed leak after stent treatment and removal.

Two patients had persistent gastrocutaneous fistulas. In patient 1, the fistula eventually resolved with stent therapy. Patient 6 was treated for her chronic fistula with several episodes of endoscopic stent placement and Tisseal fibrin glue (Baxter, Deerfield, IL) injections. After 7 months, the patient underwent elective surgical revision of her RYGB with conversion of the GJ anastomosis to an EJ anastomosis and a temporary cervical esophagostomy. She recovered, the diverting esophagostomy was reversed, and the fistula did not recur. The mean follow-up was 4.8 months (range 1–16), with

leak closure by stents in 5 (83%) of the 6 patients. The median time from stent placement to leak closure as demonstrated on UGI contrast study was 44 days. Discussion The incidence of anastomotic leaks after RYGB is low but is associated with major morbidity and mortality. In a large prospectively collected multicenter study of 3018 consecutive patients [2], 63 (2.1%) developed an anastomotic leak. Of

598

C. A. Edwards et al. / Surgery for Obesity and Related Diseases 4 (2008) 594 – 600

Table 2 Operative data Pt. No.

Leak type

Stent OR time (min)

Stent duration (d)

Stent revisions (n)

Complications

Stent leak resolution (d)

1

15

First, 2 Second, 2 0 0 2

467

20 19 6

First, 46 Second, 295 18 43 30

Migration

2 3 4

Acute Chronic Acute Acute Acute

5 6

Acute Chronic

52 23

28 270

0 5

Migration None Migration, with stents passed per rectum Migration Migration

38 44 37 45 Revisional surgery

Pt. No. ⫽ patient number; OR ⫽ operating room.

these 63 patients, 40 required operative treatment, which included drainage of abdominal fluid collections, washout of the abdominal cavity, placement of closed suction drains in 17 (43%), and repair of the anastomosis in 16 (40%). Of these 40 patients, 4 died (10%). Of the 63 patients, 23 were treated nonoperatively by maintenance of a drain placed at surgery, antibiotics and nothing by mouth, with no mortality. The postoperative complication rate for all 63 patients was 59%. The major complications after an anastomotic leak included GJ stricture (13%), gastrogastric fistula (10%), small bowel obstruction (8%), wound infection (19%), incisional hernia (10%), respiratory failure (10%), deep venous thrombosis (6%), pulmonary embolism (8%), and multisystem organ failure (16%). Patients with hypotension and oliguria were more likely to be treated operatively. All patients received specialized nutrition with parenteral nutrition in 63% and enteral nutrition alone in 37%. The use of stents for anastomotic leaks after RYGB can improve and simplify management. The stent immediately seals the leak from additional contamination of the peritoneal cavity. The patient can generally start oral feeding soon after stent placement, allowing oral nutrition and eliminating the need for costly and high-risk total parenteral nutrition. In our patient group, 5 of 6 were able to resume oral feeding within 1 week of stent placement. This is a vast improvement compared with most studies in which all patients required total parenteral nutrition or enteral tube feedings. The stents allow healing while nutrition is maintained. We believe this will reduce the costs and morbidity in this patient group. No mortality and no major morbidity occurred in our 6 patients. All acute leaks eventually closed without stricture or fistula development. These stents were used as either the only procedure or as an adjunctive modality in the overall management of GJ or EJ anastomotic leaks. The primary morbidity was that of tube migration, which improved when we used longer tubes of a larger diameter. All stents were eventually removed without morbidity. Other studies have demonstrated the efficacy of stents in the management of thoracic esophageal leaks [3–7]. Two studies have had good results with the use of partially covered, self-expanding metal stents in bariatric patients with fistula formation after RYGB [8,9]. A recent study [10] described the use of

SEPSs in 4 patients to treat anastomotic complications after gastric bypass surgery, with results similar to ours. The main disadvantage of this new modality is stent migration. Stent migration required additional endoscopic intervention to replace the stents in 4 of 6 patients. Most of the stents migrated only a few centimeters, but the migration was significant enough to result in a recurrent anastomotic leak. The morbidity of stent repositioning or replacement included the additional procedure and general anesthesia time. The length of the polyester stent appeared to be a significant factor leading to stent migration. Our initial series used smaller 16-mm-wide and 9- or 12-cm-long stents. All these stents migrated to some extent. The migration was less when replaced with 18-mm-wide and 15-cmlong stents. One patient had a stent migrate distally into the small bowel on 2 occasions; both stents passed per rectum without additional intervention. Although no complications directly related to the passed stent resulted, a theoretical risk of small bowel obstruction or perforation still exists. Therefore, surveillance for stent migration might be necessary after placement. In addition to early UGI contrast studies, occasional plain radiographs are needed to monitor for stent migration. Appropriate replacement of these migrated stents is necessary when migration occurs. These stents were designed for placement across esophageal malignant and anastomotic strictures, not across a fresh, widely patent, gastrojejunal anastomosis. The placement of endoscopic clips, laparoscopic absorbable sutures, or routine replacement at short regular intervals might be needed to minimize migration. The design of a new type of stent specifically created for bariatric surgery might eventually be required to solve the problem. We believe that if a stent can be kept in place across a gastrojejunal anastomotic leak for an appropriate duration, healing can be accomplished in most patients while allowing for simultaneous oral intake. Conclusion The use of SEPSs in the management of gastrojejunal and gastroesophageal leaks after gastric bypass surgery is a novel technique that has promise to decrease the morbidity compared with current treatments. This series has revealed

J. Alverdy / Surgery for Obesity and Related Diseases 4 (2008) 594 – 600

the safe and effective use of endoscopically placed SEPSs for acute gastrojejunal or esophagojejunal anastomotic leaks in a small series of selected patients. The data have demonstrated adequate exclusion of anastomotic leaks with simultaneous oral nutritional intake, as well as eventual closure of the leaks. The disadvantage of this stent was the high rate of migration. This modality is still investigational, and additional study is needed to determine its optimal role in managing leaks after bariatric surgery. Disclosures The authors claim no commercial associations that might be a conflict of interest in relation to this article. References [1] Baker R, Foote J, Kemmeter P, Brady R, Vroegop T, Serveld M. The science of stapling and leaks. Obes Surg 2004;14:1290 – 8. [2] Gonzalez R, Sarr MG, Smith CD, et al. Diagnosis and contemporary management of anastomotic leaks after gastric bypass for obesity. J Am Coll Surg 2007;204:47–55.

599

[3] Ott C, Rattiu N, Endlicher E, et al. Self-expanding polyflex plastic stents in esophageal disease: various indications, complications, and outcomes. Surg Endosc 2007;21:889 –96. [4] Langer FB. Management of postoperative esophageal leaks with the polyflex self-expanding covered plastic stent. Ann Thorac Surg 2005; 79:398 – 403. [5] Hunerbein M. Treatment of thoracic anastomotic leaks after esophagectomy with self-expanding plastic stents. Ann Surg 2004; 240:801–7. [6] Scheidbach S. Endoscopic treatment of thoracic esophageal anastomotic leaks by using silicone-covered, self-expanding polyester stents. Gastrointest Endosc 2005;61:897–900. [7] Gelbmann CM, Rath HC, Rogler G, Lock G, Schölmerich J, Kullmann F. Use of self-expandable plastic stents for the treatment of esophageal perforations and symptomatic anastomotic leaks. Endoscopy 2004;36:695–9. [8] Eisendrath P, Himpens J, Cadière GB, Le Moine O, Devière J. Endotherapy including temporary stenting of fistulas of the upper gastrointestinal tract after laparoscopic bariatric surgery. Endoscopy 2007;39:625–30. [9] Salinas A, Santiago E, Antor M, Salinas J. Self-expandable metal stents to treat gastric leaks. Surg Obes Relat Dis 2006;2: 570 –2. [10] Fukumoto R, Orlina J, McGinty J, Teixeira J. Use of polyflex stents in treatment of acute esophageal and gastric leaks after bariatric surgery. Surg Obes Relat Dis 2007;3:68 –72.

Editorial comment

Comment on: Management of anastomotic leaks after Roux-en-Y bypass using self-expanding polyester stents Edwards et al. present a retrospective review of their experience using self-expanding polyester stents in 6 patients after 5 open revisional bariatric procedures and 1 laparoscopic Roux-en-Y gastric bypass performed as a primary procedure. The indication for stent placement was a clinically significant leak at the gastrojejunal anastomosis, although 1 patient had leakage at a previously constructed esophagojejunostomy. As in all cases involving anastomotic or staple line disruptions in this region, the authors navigated the fine line between supportive care and expectant observation versus operative control of the leak, with drainage, abdominal washout, and enteral access for feeding. Although the report is not intended to address the specific indications for stent placement in lieu of operative management, it appears that, in many of the cases, most experienced surgeons would have viewed operative re-exploration as a maneuver to be avoided. Because the indications for, and management of, acute and chronic leaks in proximal staple lines and anastomoses are debated, this report was not intended to clarify the issues and the results do not provide any new insights in this regard. However, the results of this series are valuable, because they demonstrate the feasibility of stent

placement after leaks that develop both acutely and remotely from the initial operation, as well as those that are contained and noncontained. Although stent migration is a clear and risky complication of this approach, stent migration appeared to be a clinical nuisance and not a major issue to lend caution to the procedure. All stents were placed fluoroscopically, and the authors do not address why endoscopy was not used. Nonetheless, the report demonstrates yet another possible approach to a dreaded complication bariatric surgeons face infrequently, but painfully. Without spiraling into a typical sermon from an academic surgeon, pointing out that this is a retrospective review requiring a properly controlled trial to make meaningful conclusions, I am struck by the attractiveness of this approach in situations in which prolonged inanition is anticipated. Nothing is more difficult and frankly depressing for the patient, family, and treating physician to impose a seemingly inexorable and unpredictable period of nothing by mouth to manage a complex gastric leak. The ability to repeatedly place stents, immediately test their integrity to “seal” using oral contrast agents, and to allow the patient to take oral nutrition is a major advantage of this approach that will force surgeons to