Clinical validation of the ISGPF classification and the risk factors of pancreatic fistula formation following duct-to-mucosa pancreaticojejunostomy by one surgeon at a single center.

BACKGROUND: Postoperative pancreatic fistula remains a troublesome complication after pancreatoduodenectomy (PD), and many authors have suggested factors that affect pancreatic leakage after PD. The International Study Group on Pancreatic Fistula (ISGPF) published a classification, but the new criteria adopted have not been substantially validated. The aims of this study were to validate the ISGPF classification and to analyze the risk factors of pancreatic leakage after duct-to-mucosa pancreatojejunostomy by a single surgeon. METHODS: All patient data were entered prospectively into a database. The risk factors for pancreatic fistula were analyzed retrospectively for 247 consecutive patients who underwent conventional pancreatoduodenectomy or pylorus-preserving pancreatoduodenectomy between June 2005 and March 2009 at the Samsung Medical Center by a single surgeon. Duct-to-mucosa pancreatojejunostomy was performed on all patients. The ISGPF criteria were used to define postoperative pancreatic fistula. RESULTS: Conventional pancreatoduodenectomy was performed in 84 patients and pylorus-preserving pancreatoduodenectomy in 163. Postoperative complications occurred in 144 (58.3%) patients, but there was no postoperative in-hospital mortality. Pancreatic fistula occurred in 105 (42.5%) [grade A, 82 (33.2%); grade B, 9 (3.6%); grade C, 14 (5.7%)]. However, no difference was evident between the no fistula group and the grade A fistula group in terms of clinical findings, including postoperative hospital stays (11 versus 12 days, respectively, p = 0.332). Mean durations of hospital stay in the grade B and C fistula groups were significantly longer than in the no fistula group (21 and 28.5 days, respectively; p < 0.001). Multivariate analysis revealed that a soft pancreas and a long operation time (>300 min) were individually associated with pancreatic fistula formation of grades B and C. CONCLUSIONS: Although the new ISGPF classification appears to be sound in terms of postoperative pancreatic leakage, grade A fistulas lack clinical implications; thus, we are of the opinion that only grade B and C fistulas should be considered in practice. A soft pancreatic texture and an operation time exceeding 300 min were found to be risk factors of grade B and C pancreatic fistulas.