ABSTRACT
BACKGROUND: Pancreatic reconstruction following pancreaticoduodenectomy (PD) is still debatable even for pancreatic surgeons. Ideally, pancreatic reconstruction after PD should reduce the risk of postoperative pancreatic fistula (POPF) and its severity if developed with preservation of both exocrine and endocrine pancreatic functions. It must be tailored to control the morbidity linked to the type of reconstruction. This study was to show the best type of pancreatic reconstruction according to the characters of pancreatic stump. METHODS: We studied all patients who underwent PD in our center from January 1993 to December 2015. Patients were categorized into three groups depending on the presence of risk factors of postoperative complications: low-risk group (absent risk factor), moderate-risk group (presence of one risk factor) and high-risk group (presence of two or more risk factors). RESULTS: A total of 892 patients underwent PD for resection of periampullary tumor. BMI >25 kg/m2, cirrhotic liver, soft pancreas, pancreatic duct diameter <3 mm, and pancreatic duct location from posterior edge <3 mm are risk variables for development of postoperative complications. POPF developed in 128 (14.3%) patients. Delayed gastric emptying occurred in 164 (18.4%) patients, biliary leakage developed in 65 (7.3%) and pancreatitis presented in 20 (2.2%). POPF in low-, moderate- and high-risk groups were 26 (8.3%), 65 (15.7%) and 37 (22.7%) patients, respectively. Postoperative morbidity and mortality were significantly lower with pancreaticogastrostomy (PG) in high-risk group, while pancreaticojejunostomy (PJ) decreases incidence of postoperative steatorrhea in all groups. CONCLUSIONS: Selection of proper pancreatic reconstruction according to the risk factors of patients may reduce POPF and postoperative complications and mortality. PG is superior to PJ as regards short-term outcomes in high-risk group but PJ provides better pancreatic function in all groups and therefore, PJ is superior in low- and moderate-risk groups.
