Prediction of pancreatic anastomotic failure after pancreatic head resection using preoperative diffusion-weighted MR imaging.

ABSTRACT

PURPOSE: To determine whether the preoperative pancreatic apparent diffusion coefficient (ADC) can be used to predict the development of postoperative pancreatic anastomotic failure (PAF). MATERIALS AND METHODS: We retrospectively examined the cases of 79 patients who underwent pancreatic head resection between January 2010 and October 2013. The patients underwent 1.5-T MR imaging including diffusion-weighted imaging before surgery. The main pancreatic duct diameter (MPD), the pancreatic parenchymal thickness (PT), and the ADC of the pancreatic remnant parenchyma were measured. Two radiologists blinded to the patients' outcomes performed the measurements. The imaging parameters were compared between the patients who developed PAF and those who did not. The cut-off ADC for the development of PAF was calculated with a receiver operating characteristic analysis. RESULTS: The imaging parameters were highly correlated between the two observers. The MPD and PT did not differ significantly among the patients. The mean pancreatic ADCs were significantly higher in the patients with PAF than in those without PAF. An ADC higher than 1.50 × 10(-3) mm(2)/s (Az = 0.719, observer-1) or 1.35 × 10(-3) mm(2)/s (Az = 0.752, observer-2) was optimal for predicting the development of postoperative PAF. CONCLUSION: Measuring the preoperative non-tumorous pancreatic ADC may be useful for the prediction of a postoperative PAF.