The value of gadoxetic acid-enhanced and diffusion-weighted MRI for prediction of grading of pancreatic neuroendocrine tumors

Background: Parenchyma-preserving resection for the treatment of benign pancreatic neuroendocrine tumors (NETs) has been tried, and preoperative prediction of benign pancreatic NET is important. Recently, diffusion-weighted imaging (DWI) of abdomen magnetic resonance imaging (MRI) has been used to characterize benign and malignant tumors and DWI might be helpful in prediction of benign pancreatic NETs. Purpose: To evaluate the value of gadoxetic acid-enhanced MRI and DWI in predicting benign pancreatic NETs for determination of parenchyma-preserving resection. Material and Methods: Our ethics committee approved this study with a waiver of informed consent given its retrospective design. We searched radiology and pathology databases from November 2010 to July 2012 to identify patients who underwent surgery for pancreatic NETs (<4 cm). Twenty patients in the benign group and 14 patients in the non-benign group were included in this study. Two radiologists analyzed the morphologic features, signal intensity on MR images including DWI (b = 800), and dynamic enhancement pattern of the tumors with consensus. The tumor-to-parenchyma ratio and tumor apparent diffusion coefficients (ADCs) were quantitatively assessed. Results: The benign pancreatic NETs were more often round (7/20, 35%) or ovoid (13/20, 65%) in shape and less hypovascular on the arterial phase (3/20, 15%) than were the non-benign pancreatic NETs (1/14, 7.1% and 5/14, 35.8%; 7/14, 50% respectively; P < 0.05). Main pancreatic duct dilatation by tumors was demonstrated only in non-benign pancreatic NETs (4/14, 28.4%; P = 0.021). ADC values and ratios were significantly different between benign pancreatic NETs (mean, 1.48 x 10(-3) mm(2)/sec, 1.11 +/- 0.25, each) and non-benign pancreatic NETs (mean, 1.04 x 10(-3) mm(2)/sec, 0.74 x 0.13, each) (P < 0.01). Other qualitative and quantitative analyses between benign and non-benign pancreatic NETs were not significantly different (P > 0.05). Conclusion: Abdominal MRI with DWI may be useful for differentiating benign pancreatic NETs from non-benign pancreatic NETs, which might be helpful for determination of parenchyma-preserving resection.