AJR:193, September 2009 W175 as “drain output of any measurable volume of fluid on or after postoperative day 3 with an amylase content greater than three times the serum amylase activity” [11]. The overall rate of pancreatic fistula after pancreaticoduodenectomy ranges from 17% in patients in whom the pancreatic remnant has a hard consistency [12] and exocrine pan- creatic function and pancreatic juice output are impaired [13] to 25% in cases in which the pancreas is soft [1], that is, the parenchy- ma is normal [13]. Soft pancreatic texture [1, 3, 13] and small pancreatic duct size [3] are the most important preoperative risk factors for the development of pancreatic fistula. The diagnosis of pancreatic fistula usually is made an average of 7 days after pancreati- coduodenectomy [5, 14, 15]. Pancreatic fistula diagnosed with repeated assays of pancreatic enzymes in peripancreatic fluid drainage [11, 14, 16] is recognized in only 70–75% of cases [2, 5, 17]. The peripancreatic drains routinely Utility of CT in the Diagnosis of Pancreatic Fistula After Pancreaticoduodenectomy in Patients with Soft Pancreas Onorina Bruno 1,2 Giuseppe Brancatelli 3,4,5 Alain Sauvanet 6 Marie Pierre Vullierme 1,2 Vincent Barrau 1 Valérie Vilgrain 1,2,7 Bruno O, Brancatelli G, Sauvanet A, Vullierme MP, Barrau V, Vilgrain V 1 Université Paris 7 Denis Diderot, Paris F-75018, France. 2 AP-HP, Department of Radiology, Hôpital Beaujon, Clichy, France. 3 Department of Radiology, Università di Palermo, Via Villaermosa 29, Palermo, 90139, Italy. Address correspondence to G. Brancatelli (gbranca@yahoo.com). 4 University of Pittsburgh School of Medicine, Pittsburgh, PA. 5 Radiology Unit, La Maddalena Hospital, Palermo, Italy. 6 Service de Chirurgie Digestive, Hôpital Beaujon, Clichy F-92100, France. 7 INSERM, U773, Centre de Recherché Biomédicale Bichat-Beaujon, Paris F-75018, France. Gastrointestinal฀Imaging฀•฀Original฀Research WEB This is a Web exclusive article. AJR 2009; 193:W175–W180 0361–803X/09/1933–W175 © American Roentgen Ray Society P ancreaticoduodenectomy is safe in the management of various malignant and benign diseases of the pancreatic head and peri- ampullary region. Although the mortality rate has decreased to approximately 1–2% at high-volume centers, the morbidity rate rang- es from 30% to 50% [1–4]. The two most frequent complications of pancreaticoduo- denectomy are delayed gastric emptying and pancreatic fistula. Pancreatic fistula after pancreaticoduo- denectomy is a serious complication result- ing in prolonged hospital stay, increased costs, readmission, and a mortality rate of 3–9% [5, 6]. Efforts to reduce the incidence of pancreatic fistula have included definition of risk factors [1, 7], improvement of surgical technique [8, 9], and perioperative adminis- tration of somatostatin and its analogues [1, 10]. The International Study Group on Pan- creatic Fistula has defined pancreatic fistula Keywords: CT, pancreas, pancreaticoduodenectomy, pancreatic fistula, soft pancreas DOI:10.2214/AJR.08.1800 Received September 9, 2008; accepted after revision December 29, 2008. OBJECTIVE. The purpose of this study was to evaluate the sensitivity and specificity of routine performance of CT on postoperative day 7 in patients at high risk of pancreatic fistula after pancreaticoduodenectomy. MATERIALS฀AND฀METHODS. Two radiologists analyzed images from CT examina- tions of 50 patients with soft pancreas 7 days after pancreaticoduodenectomy. Pancreatic fistu- la was defined at CT as a fluid collection close to the pancreaticogastric or pancreaticojejunal anastomosis. Clinicobiologic criteria for the diagnosis of pancreatic fistula were drain output of any measurable volume of fluid on or after postoperative day 3 that had an amylase content more than three times the serum amylase activity. The final diagnosis of pancreatic fistula was rendered on the basis of clinicobiologic data at hospital discharge or at first readmission. RESULTS. At hospital discharge or at first readmission, 27 of 50 patients (54%) had a pancreatic fistula. On postoperative day 7, 30 patients (60%) had a total of 51 fluid collections, and CT showed a fluid collection close to the pancreaticogastric or pancreaticojejunal anasto- mosis in 21 of 51 cases. CT had a sensitivity of 63% (17/27 patients) and a specificity of 83% (19/23 patients) for the diagnosis of pancreatic fistula with four false-positive and 10 false- negative findings. The diagnosis of pancreatic fistula on the basis of clinicobiologic criteria on postoperative day 7 was made in 22 of 27 patients (81%), whereas five cases were false- negative. Four of these patients had CT evidence of pancreatic fistula. CONCLUSION. In patients at high risk who have undergone pancreaticoduodenectomy, systematic postoperative CT may be proposed as a complementary tool in the diagnosis of pancreatic fistula, particularly for detection of clinically occult pancreatic fistula. Bruno et al. CT of Pancreatic Fistula Gastrointestinal Imaging Original Research