Early gastrogastric fistula after Roux-en-Y gastric bypass: successful fistula treatment with self-expandable endoscopic stent

Abstract

A 25-year-old patient underwent laparoscopic Roux-en-Y gastric bypass surgery with an initially uneventful postoperative course. Two weeks postoperatively, the patient presented with acute abdominal pain. CT scan revealed a gastrogastric fistula from the gastric pouch to the gastric remnant. Laparoscopic drainage was performed, and intraoperative endoscopy confirmed a large gastrogastric fistula. Due to intense adhesions between pouch and remnant, a closure by suture of the fistula was not possible. The fistula was initially treated with a fully covered metal stent. After multiple stent migrations despite clip attachment to the mucosa, the stent was changed to a partially covered metal stent. Fistula healing progress was documented every 2 weeks. After 10 weeks of stent treatment, fistula closure was accomplished.

In conclusion, early fistula from the gastric pouch to the gastric remnant is a rare complication and can be managed with endoscopic stent placement.

Background

Laparoscopic Roux-en-Y gastric bypass (LRYGB) is one of the most commonly performed metabolic procedures worldwide.¹ Several studies have shown its efficacy and safety.^{2 3} Nevertheless, short-term complications such as bleeding, anastomotic leakage or perforation occur and may impose a life-threatening condition to the patient.^{4 5} Anastomotic or staple line leakage and marginal ulcers may develop into a gastrogastric fistula, a rare complication that commonly appears months to years after gastric bypass surgery.^{6 7} In this report, we present a case with a very early complication of a fistula from the gastric pouch to the gastric remnant.

Case presentation

A 25-year-old patient was admitted to the bariatric centre due to morbid obesity with an initial body mass index of 44.8 kg/m². Besides vitamin deficiencies (vitamin D: 26 nmol/L (reference: >75 nmol/L), folic acid 4.5 nmol/L (reference >8.8 nmol/L), insulin resistance (homoeostatic model assessment 9.2 (reference <2.0) and a suspicion of non-alcoholic steatohepatitis (alanine transaminase 106 U/L (reference <50 U/L)), there were no relevant obesity-related comorbidities or other health conditions. Preoperative upper endoscopy revealed a gastritis, Helicobacter pylori was not found histologically.

LRYGB with a 150 cm Roux limb and 80 cm biliary limb was performed (figure 1). Gastric pouch was transected using one transverse and four longitudinal 45 mm linear staplers (Powered Echelon Flex, Ethicon, J&J Medical Devices, USA). After splitting of the greater omentum, an antecolic end-to-side gastroenterostomy was performed using 2/3 of a 45 mm linear stapler, closing the defect with a running barbed suture (Stratafix 3–0, Ethicon, J&J Medical Devices, USA). A stapled side-to-side jejunojejunostomy was formed with a 45 mm linear stapler as well. Intraoperative methylene blue test was conducted to exclude leakage. Finally, the mesenteric defects were closed with a non-absorbable suture and drainage was placed.

Figure 1.

Schematic illustration of Roux-en-Y gastric bypass. AL, alimentary limb; BPL, biliopancreatic limb; CL, common limb.

Barium swallow tests on postoperative day (POD) 1 showed normal passage to the Roux limb without signs of leakage and food intake was started subsequently (figure 2A). The drainage was removed on POD 3 and the patient could be discharged home in good condition on POD 4. On POD 7, the patient was readmitted to hospital due to an episode of epigastric pain.

Figure 2.

(A) Barium contrast X-ray. A: Roux-en-Y gastric bypass postoperative day 1. (B) Passage of contrast agent over the fistula to the gastric remnant (arrow). (C) Bridged fistula and anastomosis with 23 cm metal stent.

Investigations

Blood tests showed leucocytosis of 11.5×10⁹/L (reference 3.2–10×10⁹/L) and C reactive protein (CRP) of 100 mg/L (reference <5 mg/L). Thoracoabdominal CT showed a thickened wall of the gastric remnant without any other abnormalities (figure 3A). Epigastric pain completely resolved, and the patient was discharged the next day. On POD 14, the patient presented again with similar symptoms and elevated temperature of 39.3°C. A leucocytosis of 11.5×10⁹/L and a CRP of 173 mg/L were found. CT revealed a 30×40 mm perisplenic fluid collection and contrast medium in the gastric remnant (figure 3B).

Figure 3.

CT, coronary view at postoperative day 7 (A) and 14 (B). Arrows indicate gastric remnant.

Due to persistent abdominal pain and fewer laparoscopy was performed. Local contamination and peritonitis was excluded. The fistula could not be visualised due to dense adhesions from the gastric pouch to the gastric remnant. However, intraoperative gastroscopy confirmed the gastrogastric fistula distally to the Z-line (figure 4A), and even intubation of the gastric remnant with the endoscope was possible.

Figure 4.

Endoscopic view of the gastrogastric fistula (arrows). (A) Intraoperative endoscopy, (B) after 10 days, (C) after 6 weeks (D) after 10 weeks of stent treatment.

Treatment

Perihepatic and perisplenic drains were placed and the gastric remnant was relieved with a percutaneous gastrostomy tube. The next day, a fully covered, 10 cm metal stent (Niti-S Stent, TaeWoong Medical, South Korea) was placed in the pouch to cover the fistula. The gastroenterostomy was not bridged to avoid erosion. Parenteral feeding was established, because enteral feeding through a nasojejunal tube was not tolerated by the patient. The pain decreased and the blood tests improved within a few days to leucocytes of 6.7×10⁹/L and CRP 78 mg/L. However, the fistula was still apparent in barium swallow tests on POD 18 (figure 1B). Endoscopy was repeated, revealing a remaining fistula of 1.5 cm diameter (figure 4B). Thus, the stent was extracted and replaced with a fully covered, 23 cm metal stent (Niti-S Stent) bridging from the oesophagus over the fistula and the gastroenterostomy to the jejunum. Consecutively, barium swallow tests did not show any leakage through the fistula (figure 1C) and liquid food intake was established. Six days after the stent replacement, the patient developed abdominal pain. CT scan showed stent dislocation (figure 5A). Upper endoscopy was performed and the stent was repositioned and fixated with a clip (Novaclip R3, 11 mm, Vytil, France) to the oesophageal mucosa. Finally, the patient was discharged in good condition with the stent in place.

Figure 5.

CT overview after placement of a fully covered 23 cm oesophageal metal stent. (A) Stent position slightly distally dislocated, bridging the gastrojejunostomy but not the gastrogastric fistula. (B) Distal stent dislocation. Arrows indicate stent.

Two weeks later, the patient presented again with abdominal pain. The long oesophageal stent dislocated distally (figure 5B) and had to be repositioned and fixated once again in the same technique. Meanwhile the fistula shrunk to 3 mm in upper endoscopy (figure 4C).

The same event occurred another 2 weeks later, so that stent was replaced with a partially covered, 10 cm metal stent (Evolution Oesophageal Controlled-Release Stent-Partially Covered, Cook Medical, USA). After further shrinking of the fistula, the stent was finally removed 10 weeks after revisional laparoscopy and the first stent placement. Control endoscopy did not reveal any further fistula opening (figure 4D). Due to the several stent dislocations and consecutive repositioning, food intake was limited and temporary parenteral feeding was established via peripherally inserted central catheter.

Outcome and follow-up

After stent removal and fistula healing, food intake was gradually increased and parenteral feeding could be stopped. Outpatient visits were held 2 and 4 months after stent removal, where no gastrointestinal complaints were reported and food intake was unproblematic.

Discussion

In this case report, we present a rare short-term complication after Roux-en-Y gastric bypass. Gastrogastric fistulas occur in up to 1%–6% after open Roux-en-Y gastric bypass, but commonly with a latency of months or years and are usually detected by weight regain. The cause is often a staple line failure in non-transsected pouch formation or a consequence of peptic ulcer in transsected pouches from the remnant. In laparoscopic RYGB, the prevalence of gastrogastric fistulas has become rare.^7–9 In a gastrogastric fistula of this size, where intubation with an endoscope is possible, incomplete division of the stomach during pouch creation must be excluded. In multiple reviews of the initial operation video, complete division could be confirmed. Furthermore, the first postoperative CT scan showed a continuous staple line up until the angle of his without oral contrast in the gastric remnant.

There are two main hypotheses how this early postoperative gastrogastric fistula occurred. A small, primarily unapparent staple line leakage at the cranial end of the gastric pouch or the remnant that has led to an abscess and subsequently to the creation of the fistula within the first two postoperative weeks. Alternatively, a thermic lesion between the Z-line and the gastric pouch might have perforated over time and then created the fistula. In both cases, initial postoperative barium swallow and blood tests may show normal findings. However, the first CT scan 1-week postoperatively did not show any fluid or air in the area next to the gastric pouch and gastric remnant. A pre-existing condition such as a gastric ulcer is unlikely to be responsible for the postoperative course, since medical record was inconspicuous and preoperative upper endoscopy excluded a gastric ulcer.

When the diagnosis was established during the intraoperative upper endoscopy, revealing the localisation and the extent of the fistula, treatment options were debated. Since there were many adhesions from the gastric pouch to the gastric remnant and the tissue appeared dense and yet fragile, fistula treatment with linear stapling as described in cases of late gastrogastric fistula seemed unsafe due to poor tissue quality.^{7 10} Therefore, only laparoscopic lavage and drainage was done in the acute setting. The resection of the pouch and the remnant fundus with reconstruction by oesophagojejunostomy seemed inappropriate in the acute inflammatory situation. In endoscopic treatment of leakage and fistula, both stenting and endoscopic vacuum therapy (EVT) have shown positive results in various cases.^11–14 Due to the large fistula size, EVT was primarily not a valid option and stenting was preferred.

Oesophageal stenting in case of leakage and fistula is an established therapy option, nevertheless, complications are not uncommon. Stent migration, which occurred several times in this case, occurs in up to 32%.¹⁵ The Niti-S stent has shown satisfactory results in terms of prevention of migration in patients with cancer, nevertheless the long stent used in our case report migrated several times.¹⁶ The use of metal clips to prevent migration has also been verified in a randomised controlled trial, but did not prevent migration neither.¹⁷ Finally, after changing the fully covered metal stent to a partially covered stent, migration has not occurred anymore, since the tendency to migrate is lower in this stent type.¹⁸ Furthermore, serious complications of oesophageal stenting, like the development of an aorto-oesophageal fistula, may occur. Early diagnosis and therapy is crucial.^{15 19}

In conclusion, early fistula from the gastric pouch to the gastric remnant can be managed with endoscopic stent placement. Diagnostic laparoscopy and drainage might be indicated to diagnose and control local contamination. Stent migration is common and must be excluded in case of any suspicious symptoms.

Learning points.

• Early gastrogastric fistula after Roux-en-Y gastric bypass can be treated with endoscopic stent placement.

• Diagnostic laparoscopic may be used to diagnose and control local contamination.

• Stent migration is common and must be excluded at a low threshold in case of any suspicious symptoms.

Ethics statements

Patient consent for publication

Obtained.