Risk factors for pancreatic anastomotic leakage: the significance of preoperative dynamic magnetic resonance imaging of the pancreas as a predictor of leakage.

BACKGROUND

The histologic degree of pancreatic fibrosis can be assessed preoperatively by using the time-signal intensity curve (TIC) of the pancreas obtained from dynamic magnetic resonance imaging.

STUDY DESIGN

To identify risk factors for postoperative pancreatic anastomotic leakage and to assess the impact of pancreatic TIC on this complication, 89 patients who underwent a pancreatic head resection with an end-to-side pancreaticojejunostomy between December 1998 and August 2005 were retrospectively reviewed. The pancreatic TIC profiles were classified into 3 types: type I, indicating a normal pancreas without fibrosis; and types II and III indicating fibrotic pancreas.

RESULTS

Pancreaticojejunal anastomotic leakage occurred in 14 patients (16%). In a univariate analysis, pancreatic texture (hard, 3% versus intermediate, 20% versus soft, 23%; p = 0.046), pancreatic duct size (> 3 mm, 8% versus <or= 3 mm, 25%; p = 0.037), and pancreatic TIC (types II, III, 3% versus type I, 25%; p = 0.006) were notably associated with pancreatic anastomotic leakage. In a multivariable analysis, pancreatic TIC (odds ratio [OR], 9.58; 95% CI, 1.1 to 91.7) was the only marked independent predictor of postoperative pancreatic leakage. A subanalysis of 52 patients with type I pancreatic TIC demonstrated hemoglobin A1c (odds ratio, 9.81; 95% CI, 1.2 to 127.9) to be a notable predictor of leakage and pancreatic leakage developed in diabetic patients with a high hemoglobin A1c concentration (> 6.0%) than in those with a normal hemoglobin A1c level.

CONCLUSIONS

Pancreatic TIC from dynamic MRI provides reliable information for predicting risk of pancreatic anastomotic leakage after pancreatic head resection. Especially in patients with type I pancreatic TIC, the presence of uncontrolled diabetes is considered a primary risk factor for postoperative pancreatic leakage.