Introduction

Bariatric surgery is currently considered the best long-term treatment for morbid obesity. The Roux-en-Y gastric bypass (RYGB) is a commonly performed procedure that achieves significant long-term weight loss in most patients [1]. Despite the dramatic decrease in the incidence of postoperative complications during the past decade, morbidity and mortality associated with RYGB are still of concern. Symptomatic gastrogastric fistula (GGF) occurs in up to 6% of RYGB [2,3,4,5]. GGF is an abnormal communication between the gastric pouch and the excluded gastric remnant. Although GGF may be treated conservatively, patients with symptomatic GGF will require technically challenging revisional procedures.

The purpose of our study was to summarize our experience in the diagnosis and surgical management of GGF in a series of 1273 consecutive patients who underwent divided laparoscopic RYGB at three institutions.

Materials and Methods

We performed a retrospective chart review of 1273 consecutive patients who underwent divided RYGB between January 2004 and November 2014. The local Research Committee granted its approval of the study protocol. Data were obtained from the surgical, clinical, endoscopic, and radiologic reports. The patients’ demographics, surgical outcomes, and significant complications are summarized in Table 1. Preoperative upper gastrointestinal endoscopy was routinely performed, including screening for Helicobacter pylori infection. When positive, patients were treated preoperatively with 14 days of omeprazole, clarithromycin, and amoxicillin. Postoperatively, all patients were given proton pump inhibitors (PPIs) for 90 days. The patients were routinely reviewed at 1, 3, 6, and 12 months postoperatively and yearly thereafter. Follow-up visits included weight measurement, thorough anamnesis, physical exam, and laboratory screening for blood nutritional deficiencies.

Table 1 Patient characteristics demographics, surgical outcomes, and significant complications
Full size table

The Bypass Technique

The operative technique used was similar to that described by Schauer et al. [6]. All RYGB procedures were done laparoscopically. A 20–30-ml gastric pouch was created. A side-to-side, gastrojejunal anastomosis was then confectioned using 30-mm linear cartridge with a 150-cm Roux limb. The opening was closed using running, 2.0, absorbable sutures. The gastrojejunal anastomosis was tested for leaks using methylene blue injected through a nasogastric tube. A side-to-side, jejuno-jejunal anastomosis was then performed, using two 45-mm linear cartridges. At postoperative day (POD) 1 or 2, all patients underwent an upper gastrointestinal contrast study to test for leaks or obstructions. Patients were scheduled for systematic postoperative visits at 1, 3, 6, and 12 months and annually thereafter.

Diagnosis of the Fistula

As part of our approach to weight regain or pain after RYGB, all patients had both a computerized tomography (CT) scan and an upper GI endoscopy.

The Fistula Repair Technique

Adhesions from the previous surgery were completely released using sharp dissection and LigaSure® (ValleyLab, Inc., Boulder, Colorado, USA). Dissection delimitated the gastric pouch, the gastrojejunal anastomosis, and the gastric remnant. A fibrotic zone was visualized between the gastric pouch and the gastric remnant with the help of the Goldfinger dissector (Blunt Dissectors, OBTECH Medical SARL, Johnson & Johnson, Le Locle, Switzerland).

Depending on the operative findings, the GGF was classified as follows:

  1. Type 1:

    located in the higher part of the gastric pouch more than 1 cm above the gastrojejunal anastomosis

  2. Type 2:

    located in the lower part of the gastric less than 1 cm from the gastrojejunal anastomosis

In type 1 GGFs, simple resection of the gastric pouch tract was performed using reinforced (GORE® SEAMGUARD® Bioabsorbable Staple Line Reinforcement, W.L. Gore & Associates, Elkton, MD, USA) linear stapling (Tristaple® technique, black cartridges, Medtronic, Minneapolis, MN, USA) using an orally inserted, 36-French calibration tube. The other end of the fistula, with the gastric remnant, was also sleeved using the same technique. Reducing the volume of the gastric remnant avoids retained gastric antrum syndrome (RGA) [7]. Omentoplasty was also performed to maintain separation of the gastric pouch and gastric remnant.

In type 2 fistulas, en bloc resection of the gastrojejunal anastomosis was done. A new, side-to-side, gastrojejunal anastomosis was then created using a 30-mm purple cartridge of the same endoGIA stapler. The opening was closed using running, 2.0, absorbable sutures.

Regardless of the procedure, a leak test with methylene blue was always performed by injection through the orogastric tube.

A closed, suction drainage system was always placed next to the gastric pouch and brought out through the left subcostal port site. No gastric drainage was used.

Statistical Analysis

Categorical variables are reported as frequencies (%). Continuous variables are expressed as the mean and range. Weight loss after revisional surgery for GGF was analyzed using Student’s paired t test. A p value of less than 0.05 was considered statistically significant.

Results

Between January 2004 and November 2014, 1273 consecutive patients had divided RYGB, of whom 15 (1.18%) were diagnosed with symptomatic GGF postoperatively. The patients’ demographics, including height, pre- and postoperative weight, body mass index, follow-up, and surgery, are summarized in Table 2. Nine patients (60%) were women, and six (40%) were men. The mean age was 38 years (range, 27–51). The mean preoperative weight was 115 kg (range, 89–145). The mean body mass index (BMI) was 38.1 kg/m2 (range, 31–52).

Table 2 Characteristics, BMI, follow-up, and surgery date of patients with gastrogastric fistulas
Full size table

The presenting symptoms that led to a diagnosis of GGF were nausea (13/15), weight regain (12/15), epigastric pain (11/15), gastroesophageal reflux disease (GERD) (6/15), and diarrhea (2/15). Six patients (40%) previously had an endoscopically confirmed marginal ulcer. In the subset of patients with weight regain (80%), the mean percentage of excess weight loss (% EWL) at 1 year was 27.79% (range, 0–59.8). The % EWL pattern is summarized in Table 3.

Table 3 Characteristics, estimated weight loss, follow-up, and surgery
Full size table

The upper GI endoscopy showed the GGF in only 11 patients (73.3%). However, all patients had their GGF confirmed by IV and oral-enhanced CT scan, which demonstrated an abnormal presence of contrast material in the gastric remnant.

The mean time interval from the initial RYGB to GGF diagnosis was 28 months (range, 22–62). All patients required revisional surgery either because of refractory symptoms or because of weight regain. All revisional procedures were performed laparoscopically with no conversion to laparotomy. The mean follow-up period was 36 months (range, 24–72). Table 4 summarizes the type of surgery performed in each patient.

Table 4 Type of intervention by patient with EWL and TWL after revision
Full size table

Mortality was 0%. Two patients experienced postoperative complications (13.3%). The first was upper gastrointestinal bleeding requiring transfusion of three units of packed red blood cells. The second patient experienced postoperative pneumonia treated by antibiotics. None of the patients required reoperation. The mean hospital stay was 5.2 days (range, 3–10).

At 1 month postoperatively, all patients were clinically asymptomatic, and repeat endoscopy confirmed the absence of anomalies in all patients.

At 1-year follow-up after the surgery, the mean % EWL was 73.57% (range, 59.8–86.7) (Table 3) and the mean % TWL was 31.32% (range 20.8–37.5) (Table 4).

The mean BMI loss was nine points (1–17 kg/m2). The weight loss after the revisional surgery was statistically significant (p < 0.05), and the GGF was cured in all the cases.

Revision of the gastric pouch in RYGB allowed improvement of diabetes with respect to the number and doses of medications and the treatment in 77.7% (7/9) of patients with diabetes relapse after bypass.

No patient was lost to follow-up. At the last follow-up (mean 34 months, range 20–60 months), all the patients were asymptomatic with no use of PPI.

Discussion

Gastrogastric fistula (GGF) is an uncommon complication after RYGB for morbid obesity [7], with an incidence of 1 to 6% [5]. The incidence of a GGF is underestimated in the literature, as some patients might have an asymptomatic fistula, others might not have completed the diagnostic studies, and others are lost to follow-up.

In our retrospective series that included 1273 patients, the rate of documented GGF after divided RYGB was 1.18%. Most studies report a rate of GGF around 1% [8]; some have reported higher [5] or lower rates [9].

Recent publications reported postoperative confirmed leak or suspected micro leak, as well as marginal ulcer, as the main causes of GGF [7]. Cucchi et al. [5] reported a rate of 83% for documented leaks in patients diagnosed with GGF after RYGB. On the other hand, the incidence of GGF combined with marginal ulcers varies among studies and ranges from 0.6 to 16% [10, 11]. Preoperative H. pylori infection or use of NSAIDs may contribute to the postoperative development of marginal ulcer [4].

In our series, we did not find a correlation between elevated body mass index and incidence of leak formation or consequent development of gastrogastric fistulas. To increase our detection rate, we performed endoscopic and radiographic studies for patients presenting with weight loss failure, marginal ulceration, or persistent dyspeptic symptoms. In line with our experience and within 2 years postoperatively, we implemented a close follow-up plan for all our RYGB patients, reaching approximately 82% of the entire postoperative population. The follow-up schedule consisted of visits at 4 weeks and 3, 6, and 12 months after the surgery and annual visits afterward. The patients were encouraged to visit psychological and nutritional counselors trained to report any weight regain displayed by the patient.

Patients with GGF may be asymptomatic; present with nonspecific symptoms, including nausea, vomiting, bloating, pyrosis, and abdominal pain; or develop symptoms related to associated marginal ulcer (pain, hemorrhage, or perforation) or stricture (gastric outlet obstruction). They may also seek medical advice for diabetes recurrence, suboptimal weight loss, and/or weight regain, and some patients achieve acceptable, durable weight loss despite the presence of the fistula [12,13,14].

The gold standard for diagnosis of GGF is endoscopic and radiologic imaging, such as upper GI series, CT scan with oral opacification, or EGD. CT scan shows the presence of contrast or air in the gastric remnant, an indirect sign of GGF. Although it is not the test of choice for some surgeons [5], CT scan detected 100% of the GGF in our series. With UGI, we were able to detect 60% of fistulas. We have found endoscopy a valuable tool in visualizing the location of the fistula and planning for surgery. Even if the diagnosis was made by a single test, we still ran all three tests, as we believe that they are complementary to each other.

In some studies, GGF was an incidental intraoperative finding during revision bariatric surgery for failed previous intervention. Corcelles et al. [15] diagnosed 27.7% (10/36) of their cases of GGF intraoperatively.

Not all patients with GGF are candidates for surgical treatment. The initial approach should always be medical. Conservative management consists of eradication of H. pylori if present, sucralfate and high-dose PPI (40 mg twice daily), analgesics in addition to NSAIDs and smoking cessation [16]. The aim of conservative management is to reduce gastric acid secretion. This may eliminate abdominal pain and reflux symptoms and may also permit small GGFs to close spontaneously [17]. The conservative approach was successful in 20% of cases in one study; all of these fistulas were small in caliber [16]. On the other hand, medical management serves as a bridge to a more radical surgical therapy, especially in patients with complicated GGF.

Several institutions have tried endoscopic management for GGF. Endoscopic repair can be attempted prior to surgical repair [18]. It is proven safe and feasible, with short-term success but a poor long-term outcome. Multiple methods of repair have been utilized, including fibrin sealant, endoclips, and endoscopic suturing systems. However, surgery remains the most definitive approach. Recent advances in surgical management of GGF have led to a reasonable classification of GGF into type 1 and type 2, with corresponding surgical interventions (see above):

  • For type 1 GGF, excision of the fistula with preservation of the GJA, with or without calibration according to the size of pouch [4, 16].

  • For type 2 fistulas, which are usually associated with persistent marginal ulcer or anastomotic stenosis, complete revision of the gastrojejunal anastomosis with fistula excision is done [15].

The extent of resizing of the remnant stomach is controversial and varies among surgeons. Resizing the remnant may decrease the fistula rate [19,20,21], but oversizing leads to retained gastric antrum syndrome, as described by Ribeiro-Parenti et al. [9]. Some surgeons have found better results with interposition of the omentum or the jejunum to support the closure [2, 22].

Our results show that surgical management of GGF yielded good results regarding weight loss and resolution of comorbidities. These results were similar to O’Brien et al., who showed that 4 months after GGF repair, weight loss resumed and yielded favorable changes in satiety and orexigenic gut hormones [23]. Fasting and postprandial ghrelin decreased and were strongly correlated with weight loss. The insulin response to glucose also tended to be increased after GGF repair.

Although we have presented a relatively large series of patients who underwent RYGB over 10 years, a subset of whom had GGF that was treated surgically, the study was limited by its retrospective design and the absence of guidelines for the diagnosis of GGF, meaning we could only report cases of GGF that were symptomatic.

Conclusion

Roux-limb-to-gastric-remnant fistulas and gastrogastric fistulas are rare complications of RYGB procedures. However, the determination of their true incidence should have mandated systematic postoperative upper gastrointestinal endoscopy. Their pathophysiology is unknown. CT scan should always be done in case of weight regain when looking for GGF. Laparoscopic management can be performed safely and successfully in symptomatic patients.