Purpose: The aim of our study was to compare the effectiveness of MRI in predicting the location, type of ductal involvement and malignant transformation of intraductal papillary mucinous (IPM) pancreatic tumors before surgery.
Methods and Materials: Twenty four patients with histological confirmation of IPMT were included in the study. The MRI (HASTE, RARE and T1 wieghted sequences before and after gadolinium injection) performed before surgery, was retrospectively assessed. Two radiologists unawared of the initial radiological report and the final histological data independently reviewed the MR studies in order to define : the ductal type and location of the lesions, and the presence of signs of malignant transformation. Interobserver agreement was determined with weighted kappa statistics. After consensus of both radiologists, the observers MR interpretation was compared to the histologic results using weighted kappa statistics and Fisher test.
Results: Macrosopy showed combined main / collateral duct type lesions (n=17) and collateral only duct type lesions (n=7). Histologic analysis revealed adenocarcinoma (n=18), hyperplasia (n=3), and dysplasia (n=3). The lesions were located mainly in the head or uncinate process (n=16) or were diffuse or multifocal (n=2). Excellent agreement was found between the interpreters (kappa=0.90) in the evaluation of ductal involvement, good in the evaluation of lesion location (kappa=0.80), and in the diagnosis of malignant transformation (kappa=0.74). MR-histopathologic correlation was excellent in the assessment of the ductal involvement (kappa=0.90, sens=100%, spe=94%), moderate in the assessment of the tumor location (kappa=0.57, sens=87%, spe=56%), and in the diagnosis of malignant transformation (kappa=0.60, sens=83%, spe=83%). No valuable predictive sign of malignancy could be demonstrated in our study.
Conclusion: MR provides an accurate ductal type and location of the IMPT before surgery. No significant MR signs of malignancy has been identified in our group of patients.
Questions about this event email: firstname.lastname@example.org