Button jejunostomy is a novel jejunostomy creation technique first reported in 1989 [7,8,9,10]. There are some advantages of the button jejunostomy, including the simplicity of the procedure, ease of replacement, low risk of tube deviations and obstructions, patient’s aesthetic outcomes, and a high quality of life. However, a common complication is REF after button jejunostomy removal. No studies have shown the risk factors for RFF after button jejunostomy removal and the outcomes of the EPA for REF closure, until now.
In this study, the duration for button jejunostomy removal (> 12 months) and fistula length (< 15.0 mm) were found to be the independent risk factors for REF, and minimally invasive REF closure by the EPA had good outcomes.
There is no consensus on the timing of jejunostomy removal because the amount of oral intake varies depending on the patient’s primary disease stage and the presence of post-operative complications. However, according to a previous report , the risk of REF increased when the duration of jejunostomy removal was > 1 year, and the results of this study also supported the previous report. Epithelialisation of the fistula was considered to gradually form > 1 year, resulting in REF .
Since the fistula length was measured as the vertical distance between the skin and the intestinal puncture site, it was defined as the sum of the subcutaneous fat thickness plus the rectus abdominis muscle. However, the body mass index or presence of sarcopenia, which is an index of skeletal muscle mass, does not show a significant difference in the risk of REF; therefore, the fistula length should be considered as an independent risk factor. It has been reported that enterocutaneous fistulas, which close spontaneously, have long fistula tracts (> 2 cm) , and our results could be a useful cut-off value for REF in button jejunostomy with a short fistula length.
The length and thickness of the fistula are considered to be the reasons for frequent development of REF after button jejunostomy removal than after Witzel jejunostomy. The jejunostomy button used in our hospital is 24 Fr with a diameter of 8.0 mm, whereas the jejunal tube used in the Witzel jejunostomy is 10–12 Fr with a diameter of 4.0 mm. The diameter of the fistula was longer in button jejunostomy than in Witzel jejunostomy. Furthermore, in button jejunostomy, a fistula is formed vertically to the abdominal wall, but in Witzel jejunostomy, a diagonal fistula is formed; therefore, the length of the fistula in button jejunostomy is shorter than that in Witzel jejunostomy. It has been reported that enterocutaneous fistula that can be expected to spontaneously close have a diameter of ≤ 1 cm and length of ≥ 2 cm . This report supports our hypothesis that REF occurs more frequently after button jejunostomy removal. In addition, ‘free distal flow’ has been reported as an important factor in spontaneous closure . Button jejunostomy has an acute bending angle between the jejunostomy and abdominal wall compared with Wizel jejunostomy , which may interfere with the natural flow of intestinal fluid to the distal intestinal tract and lead to REF.
Spontaneous closure usually can be expected after removal of the jejunostomy tube; however spontaneous closure often requires 4–6 weeks [4, 11, 12]. Immediate fistula closure can be an effective option for patients waiting for spontaneous fistula closure that require a long period of time to close or those suffering from REF. There are some reports of minimally invasive treatments for REF, such as the use of over-the-scope clip  and fibrin glue . The success rate is reported to be approximately 50–87.5%, but the applicable fistula is a small enterocutaneous fistula with a diameter of approximately 5 mm or a low-output enterocutaneous one. It is difficult to treat a high-output fistula with a large diameter, such as REF, after removing the jejunostomy button. Although bowel resection, including the fistula, is highly curative, it can be a highly invasive treatment for patients after oesophagectomy with severe intra-abdominal adhesions, malnutrition, and poor surgical tolerance.
For the aforementioned reasons, we devised an EPA to treat the REF. This procedure is simple, has a short operative time, can be performed even under local anaesthesia, does not reach the abdominal cavity, and has the advantage of not being affected by adhesions from the previous surgery. The fistula was transected by separating and ligating it as deep as possible on the intestinal side, reducing the pressure on the intestinal side of the fistula, and closing it by causing scarring at the transected point. In this procedure, we consider that the fistula can be easily separated from the surrounding tissue as deeply as possible with good visual field, and aesthetic outcomes of the wound can be improved along the skin dividing line by creating a spindle-shaped incision of approximately 3–5 cm around the fistula.
The success rate of EPA for REF was 80%, with good results. If the patient was judged to be operable under general anaesthesia, EPA was performed under general anaesthesia in case of unexpected fistula injury or laparotomy.
This study had some limitations. First, it was retrospective and was conducted at a single centre with a small number of patients. Further studies using data from a large-scale, multicentre registry should be conducted in the future. Second, this study is limited to button jejunostomy and may not be applicable to REF in other jejunostomy creation methods. Third, fistula closure using a single EPA may not be effective in high-risk patients with cancer recurrence or poor nutrition. In addition, the definition of REF remains controversial, and in this study, the timing of performing the EPA was not unified. Finally, considering that the high incidence rate of REF after removal of the button jejunostomy (31.9%), instead of performing button jejunostomy routinely, early removal of the button jejunostomy or Witzel jejunostomy should be considered depending on the patients’ risk.