Optimal management of the pancreatic remnant after pancreaticoduodenectomy.

OBJECTIVE: The authors evaluated methods of operative management of the pancreatic remnant after pancreaticoduodenectomy. SUMMARY BACKGROUND DATA: Despite reductions in mortality after pancreaticoduodenectomy, leakage from the pancreatic remnant still may cause significant morbidity. Patients with small, unobstructed pancreatic ducts or soft, friable pancreata are at particularly high risk. Although numerous surgical techniques have been described to avoid such complications, no single method is suitable for all patients. METHODS: The authors retrospectively reviewed the medical records of 114 consecutive patients who underwent pancreaticoduodenectomy. Sixty-nine patients were men (61%) and 45 were women (39%), with median age 66 years. Underlying disease was malignant in 87 (76%) and benign in 27 (24%). Patients were divided into groups based on risk for postoperative pancreatic fistula and on the operative management of the pancreatic remnant. Sixty-eight patients underwent end-to-side pancreaticojejunostomy, 13 of whom were high risk (group 1A) and 55 of whom were low risk (group 1B). Thirty-seven patients, all high risk, had either pancreatic duct closure by oversewing (N = 19, group 2) or end-to-end pancreaticojejunal invagination (N = 18, group 3). Nine patients underwent total pancreatectomy (group 4). Morbidity related to prolonged pancreatic drainage (PPD) of greater than 20 days was determined. RESULTS: Overall incidence of PPD was 17% and caused the only death. Patients considered high risk for postoperative pancreatic fistula had a 36% incidence of PPD compared with 2% in patients considered low risk (p < 0.0001). Prolonged pancreatic drainage frequency related to the method of pancreatic remnant management was as follows: group 1A, 15%; group 1B, 2%; group 2, 79%; and group 3, 6% (p < 0.001 for group 2 vs. other groups). No serious sequelae followed PPD in 15 patients (79%); however, 4 patients required reoperation for pseudocyst or abscess drainage; one in group 1A (who died) and three in group 2. Multivariate analysis revealed that operative technique (oversewing of the pancreatic duct) and male sex were significant factors predisposing a patient to the development of PPD. CONCLUSION: After pancreaticoduodenectomy, pancreatic remnant management by end-to-side pancreaticojejunostomy appeared safe in low-risk patients. In high-risk patients, end-to-end pancreaticojejunal invagination was the safest option. Morbidity was greatest after pancreatic duct closure without anastomosis.