A 66-year-old Japanese man underwent LTG with D1 plus lymph node dissection and Roux-en-Y reconstruction for gastric cancer. His treatment course is summarized in Fig. 1. At hospitalization, his nutritional status, glucose tolerance, liver function, and lipid metabolism were all within normal limits; therefore, elective LTG was performed. Esophageal transection was performed with an endoscopic linear stapler. Then, a transorally inserted anvil (OrVil™; Covidien, Mansfield, MA, USA) was introduced into the esophagus. The operator indicated that the OrVil™ tube had reached the esophageal stump and made a small hole in the center of the esophageal stump with scissors. The tube was then extracted through the hole until the anvil reached the esophageal stump. The tube was then disconnected from the anvil by cutting the connecting thread and was removed from the abdominal cavity. Next, the jejunum was separated with a linear stapler approximately 25 cm from the Treitz ligament (Fig. 2). The elevated jejunum was retrocolically elevated. Esophagojejunostomy was then performed using a 25-mm EEA-XL stapler (Covidien, Mansfield, MA, USA). No evidence of vascular or organ injury was observed. Moreover, an intraoperative air leak test performed by immersion of the anastomosis in saline solution was performed at the end of the stapling, and no bubble escaped from the anastomosis. The operative time was 4 h 23 min, and minimal bleeding occurred.
The pathological findings were as follows: L, Post, Type 0–IIb, 54 × 36 mm, por 2 > sig, pT1a (M), UL (+), ly0, v0, pN0 (0/47), pPM0, pDM0, pStage IA.
Following our usual clinical pathway, the patient was mobile by day 1 after surgery, drinking water on day 2, and had started a liquid diet by day 3. Although he developed a fever of approximately 38 °C on day 3, no associated abdominal pain was reported, and the drainage fluid was yellow and clear. However, on day 5, the patient complained of sudden abdominal pain while walking, and the drainage fluid had become turbid. Therefore, we stopped the diet immediately. On a subsequent abdominal computed tomography (CT) scan, no effusion was found near the esophagojejunal anastomosis or under the diaphragm; therefore, we concluded that the drainage was sufficient and that the leakage was minor. Thus, we began antibiotic medication. Despite our intervention, the abdominal pain continued to worsen, and the patient could not walk; therefore, we performed oral contrast radiography with Gastrografin® on day 6. This procedure showed that the contrast medium immediately leaked from the esophageal end of the anastomosis to the pelvic space, and we diagnosed complete separation of the anastomosis (Fig. 3). We considered lavage and drainage necessary and performed emergency surgery on the same day.
During emergency laparoscopy, saliva and contaminated ascites were observed to cause a thick white collection around the esophagojejunal anastomosis. Complete separation of the esophagojejunal anastomosis was confirmed after irrigating the upper abdomen; the ends were separated by approximately 5 cm, and the remaining staples formed a B shape (Fig. 4). No evidence of color change was present in any layer of the anastomosis at the site of the elevated jejunum, suggesting that this dehiscence was unlikely due to insufficient blood perfusion. Re-anastomosis was considered too difficult in the presence of ongoing inflammation; therefore, the esophageal and jejunal openings were closed by sutures to prevent the leakage of digestive fluid and saliva. The surgery was completed by leaving drainage tubes in the esophageal end, in both sub-diaphragmatic spaces, and in the pouch of Douglas.
On day 12, 6 days after the first emergency surgery, a CT scan suggested continued peritonitis with evidence of diffuse peritoneal hypertrophy and fluid accumulation in the anterior area of the inferior duodenal flexure (Fig. 5). Therefore, a second emergency surgery was performed. We found an accumulation of ascites near the esophageal hiatus and the elevated jejunum stump. Although the esophageal end had been sutured during the previous surgery, the sutures had failed to hold, and the stump was open, with the esophagus pulled into the mediastinum, allowing saliva to leak into the abdominal cavity. However, we did not re-close the esophageal end but instead opted to leave a tube in situ to drain saliva. The elevated jejunal end was also open, but we closed this end firmly using serosa-muscle-layer sutures that introverted the mucosa. We also created an intestinal fistula to permit nutrition, which was formed at 20 cm on the anal side from the Y portion.
After the second surgery, the patient’s clinical symptoms resolved, including his abdominal pain and fever, and an improvement in his blood data (white blood cell count, 8670/mm3 on day 7 after the second surgery) was noted. We then started enteral nutrition on day 43 (Fig. 1), and although oral contrast radiography showed leakage of the contrast medium from the dehiscence into the drainage tube at 2 and 3 months after the surgery, it had stopped by 4 months postoperatively. By this time, contrast medium was observed to flow from the esophageal end to the elevated jejunum, suggesting the development of a fistula (Fig. 6). The fistula measured approximately 6 cm in length and 0.2 cm in diameter. Given the location and path of the fistula, we performed balloon dilation (Fig. 7) as a non-operative treatment every 1 or 2 weeks. Although we dilated the fistula to 0.8 cm in diameter, oral intake was not achieved, and we concluded that non-operative treatment alone would not be sufficient.
Due to the lack of a satisfactory outcome from non-operative treatment, we chose to perform esophagojejunal re-anastomosis after 7 months. A complete fistula was observed between the oesophagus and the jejunum (Fig. 8) at the point where complete separation of the esophagojejunal anastomosis had previously been observed (Fig. 4). During surgery, we exposed and taped the fistula and resected a 6-cm portion; the esophageal end showed inflammation-related hypertrophy and sclerosis, which were pulling it into the mediastinum; therefore, we dissected the esophagus beyond the dehiscence-induced inflammation and resected the sclerotic esophageal tissue using a linear stapler. We also resected the elevated jejunum stump, including the previous anastomosis site, using a linear stapler. After confirming that both the esophagus and elevated jejunum had sufficient blood perfusion and were not tense or twisted, we created a side-to-side esophagojejunal anastomosis using an Endo-GIA stapler (Covidien, Mansfield, MA, USA), overlapping both ends on the left side of the esophagus, again using an Endo-GIA stapler.
Oral contrast radiography performed 4 days after the re-anastomosis showed no signs of dehiscence (Fig. 9); therefore, oral ingestion was started 8 days after surgery. However, the patient developed an abdominal abscess secondary to a wound infection on day 9 after the re-anastomosis, progressing to endocarditis, cardiac tamponade, and bilateral pleural effusions by day 18, which caused his general condition to worsen. Because of the endocarditis and cardiac tamponade, it was suggested that extremely severe inflammation had been caused by partial injury of the pericardium during esophageal resection in the re-anastomosis. We provided ultrasound- and CT-guided abdominal, pericardial, and thoracic drainage, together with antibiotic treatment, and the patient’s general condition improved. Later, ingestion was started on day 28 after the re-anastomosis, and the patient was discharged 8 months after the initial surgery, 47 days after the re-anastomosis.
Endoscopic examination at 1 year after re-anastomosis showed no stenosis at the anastomosis site, food obstruction, or reflux esophagitis (Fig. 10). Approximately 2 years after discharge, the patient weighed 47 kg, and his ingestion had recovered to 80% of that before surgery. The patient continues to enjoy sporting activities as he previously did and attends our outpatient clinic every 3 months.