Pancreatic fistulas are a possible complication after various disorders of the pancreas. The most common reasons are inflammatory, traumatic, postoperative, or iatrogenic. Irrespective of their etiology, pancreatic fistulas may cause a whole string of complications, such as abscess, bleeding, pseudocyst, and necrosis. Therefore, an expeditious and possibly minimally invasive management is desirable. For many years, pancreatic fistulas have been managed either surgically, by pancreatic resection or internal drainage (e.g., pancreaticojejunostomy, Roux-en-Y), or conservatively, by drainage over a long period of time. However, recently, new strategies have become more popular.

Figure 1
figure 1

Gastrografin is administered via the silicon drain before intervention. A communication between the pancreatic duct and the fistula is demonstrated.

Full size image

Very encouraging have been obtained results at reducing the incidence of and managing postoperative pancreatic fistulas by using fibrin glue. In 1991, Kram et al. [11] described a method for applying fibrin glue directly to penetrating pancreatic injuries or to the tissue after pancreatic resection. Penetrating injuries were treated by completely filling the defect with a 1–2-ml fibrin glue plug. Sutures or staple lines related to the operative management of traumatic injuries were sealed intraoperatively with 4–6 ml of fibrin glue. Postoperatively, none of the patient developed pancreatic fistulas, abscesses, or pseudocysts. In contrast to our procedure, no additional internal drainage was used. In the 1980s at the same institution, 29% of patients operated on for pancreatic trauma developed fistulas [19]. At that time, no fibrin glue was used.

The first results of the endoscopic occlusion of gastrointestinal fistulas with fibrin were reported in 1990. Lange et al. [12] succeeded in closing 11 of 16 gastrointestinal fistulas using a flexible endoscope, which was advanced percutaneously along the fistula tract to seal the fistulas with fibrin glue. Until then, the endoscopic management of pancreatic fistulas usually consisted of papillotomy or transpapillary drainage of the pancreatic duct. This is a highly effective therapeutical option but may still be very time-consuming [4, 16]. Engler et al. [6], however, needed three fibrin applications through a catheter passed retrogradely into the pancreatic duct, at 2-day intervals, to close a pancreaticocutaneous fistula. We hypothesized that the combination of both fibrin glue and internal drainage of the pancreatic duct would be highly effective in the treatment of pancreatic fistulas.

Patients and methods

Over a 2-year period, eight patients with external pancreatic fistulae underwent pancreatic fistula occlusion with fibrin glue using an endoscopic technique. All cases were pancreaticocutaneous fistulas originating from the pancreatic body and tail after oncologic operations (Table 1). A pancreatic fistula was defined as pancreatic fluid discharge for >7 postoperative days. The maximal amylase concentrations in the discharged fluid ranged between 37,000 and 150,000 U/L, and the lipase concentration ranged between 44,000 and 630,000 U/L. The median discharge volume was 110 ml per day (range, 50–210). Before the intervention, fistulas had persisted for 12 days to 3 months. All patients were nil by mouth from the beginning of the treatment until the 7th day after fistula closure. Somatostatin or analogues were not used.

Table 1 Operative data for eight patients with external pancreatic fistula after oncologic upper abdominal operations
Full size table

The first step in the procedure was the percutaneous insertion of a 27 Ch silicon drain into the inflammatory peripancreatic area under sonographic or CT guidance. To achieve a stable fistula tract, this drain remained in place for ≥14 days. Two weeks after the drain was inserted, fistula closure was carried out. An endoscopic retrograde pancreatography (ERP) was performed to rule out pancreatic duct obstruction. At the same time, a prosthesis (Solopass 7 Ch, 5 cm) was inserted into the duct. To avoid stent-induced damage of the pancreatic duct, only short stents were used. In none of the cases was the fistula bridged by the prosthesis. Thereafter, a fistuloscopy was done with a bronchoscope (Olympus BF 1T10), and the fistula was rubbed with a cyto brush or pair of biopsy forceps. Any extant necrotic sequesters were then removed with the forceps. Next, a Duploject catheter was placed in the fistula. After removal of the bronchoscope, the fistula was filled to skin level with fibrin through the catheter.

Results

There was no additional morbidity or mortality due to the intervention. Immediately after fistula occlusion, fistula secretion was absent in all eight cases. An average of 8.6 ml (range, 6–10) of fibrin glue was used for each patient. In three cases, the patients complained of some abdominal pain and fever after the intervention. None of the cases required further intervention. In none of the cases was there evidence of fistula on ERP when the prosthesis was removed 4 weeks after fistula closure. None of the patients developed an abscess, recurrent fistula, or complications associated with fibrin glue, such as pancreatic duct occlusion, thrombosis, embolism, or necrosis.

Discussion

Any upper abdominal invasive procedure in the vicinity of the pancreas places the patient at risk of developing fistulas. The incidence of postoperative pancreatic fistula is reported to be between 4% and 52%, depending on the etiology and definition used [1, 9, 15, 19, 20]. External fistulas occur in ~4–6% of patients after surgical interventions on the pancreas [1] and in ~33% after trauma [9].

The success rate for the nonoperative nonendoscopic management of pancreatic fistulas is ~70% [9]. However, the average occlusion time is very long [14]. If pancreatic fistula closure does not occur after several weeks of conservative treatment, an operative approach is necessary. The success rate of surgical management is >80% [13], but a mortality rate of 8% and a morbidity rate of 13% have been reported. Therefore, minimally invasive procedures are highly desirable to shorten this time and to increase the success rate.

The series of patients with postoperative pancreaticocutaneous fistulas presented here shows that our approach of combining internal pancreatic duct drainage and external fistula occlusion yields a high success rate. Moreover, we have observed no morbidity related to the procedure so far. This approach combines two procedures that have already been used successfully in treating patients with pancreatic fistula. Occlusion of a fistula tract using fibrin glue has been reported to be successful in single cases [6, 12]. However, it seems obvious that drainage of the pancreatic duct to relieve the pressure from the fistula, and thus from the fibrin clot, would further improve these results [7, 8, 10, 17, 18, 21, 22]. All of our patients were on parenteral nutrition. The use of octreotide has been shown to be effective in the management of pancreatic fistula [2, 3, 5]. However, it has not been used in conjunction with our combined procedure.

In conclusion, our data suggest that early endoscopic management as described here is beneficial in cases of postoperative external pancreatic fistula. This therapeutic strategy is less invasive than surgery and seems to have a high success rate. In case of failure, there is no disadvantage for later operative management.