Background: The occurrence of post-operative pancreatic fistula (POPF) after pancreaticoduodenectomy is a challenging issue. The aim was to identify variables on preoperative computed tomography (CT) scan, useful to predict clinically significant POPF (grades B-C) after pancreaticoduodenectomy.
Methods: Patients presented POPF after pancreaticoduodenectomy were included from two tertiary referral centres. B/W ratio was defined by ratio of pancreas body thickness (B) to main pancreatic duct (W). The predictive parameters of POPF on CT scan were assessed with a receiving operator characteristics (ROC) curve and intrinsic characteristics.
Results: Between 2010 and 2013, 186 patients who underwent pancreaticoduodenectomy were included. POPF occurred in 25% of them, and was clinically significant in 13%. After univariate analysis, endocrine tumours (P = 0.03), main pancreatic duct size (P < 0.01) and B/W ratio (P = 0.04) were significantly associated with POPF. ROC curve showed a greater area under curve for B/W ratio (0.68) than for main pancreatic duct size (0.33). A 3.8 threshold displayed 80 and 51% for sensibility and specificity, respectively, and a negative predictive value of 94%. A B/W ratio >3.8 increased the rates of post-operative haemorrhage (odds ratio = 4.3 (1.4-13.2), P = 0.01), and reintervention (odds ratio = 3.4 (1.2-9.6), P = 0.02).
Conclusions: B/W ratio superior to 3.8 assessed on preoperative CT scan may be an easy tool to predict clinically significant POPF after pancreaticoduodenectomy.
Keywords: main pancreatic duct; pancreatic fistula; pancreaticoduodenectomy.
© 2017 Royal Australasian College of Surgeons.