Differentiating Pancreatic Cystic Neoplasms from Pancreatic Pseudocysts at MR Imaging: Value of Perceived Internal Debris

PURPOSE: To retrospectively evaluate the sensitivity and specificity of several morphologic findings that may be seen with cystic pancreatic lesions, in the diagnosis of pseudocyst at magnetic resonance (MR) imaging. MATERIALS AND METHODS: This study was institutional review board approved and HIPAA compliant. From January 1, 2005, to December 31, 2007, electronic radiology and pathology databases were searched to identify patients with pancreatic cystic neoplasms or pseudocysts who underwent pancreatic MR imaging. Twenty-two patients with cystic pancreatic neoplasms that were confirmed at surgical resection (n = 12) or endoscopic ultrasonography (US) with cystic fluid analysis (n = 10) were identified. Of 20 patients with pancreatic pseudocysts, seven had pseudocysts that were identified at pathologic resection and 13 had a clinical history of pancreatitis, with initial computed tomography (CT) revealing no pancreatic cyst and subsequent follow-up MR imaging depicting cystic lesions. Two abdominal radiologists independently and randomly evaluated each case for presence or absence of septa and internal dependent debris and for external cyst morphology on axial and coronal T2-weighted images and three-dimensional gradient-echo T1-weighted images obtained before and after intravenous contrast agent administration. Logistic regression for correlated data was used to assess the usefulness of internal debris, external morphology, and septa for differentiating cystic neoplasms from pseudocysts. RESULTS: The readers' assessments of the presence or absence of cystic debris were concordant for 40 (95%) of the 42 patients, with a kappa coefficient of 0.889, which indicated nearly perfect agreement. Thirteen (93%) of 14 lesions found to have debris by either or both readers were pseudocysts, and only one (4%) of the 22 cystic neoplasms had debris. Both readers were more likely to identify septa within cystic neoplasms than within pseudocysts; however, the difference was not significant for either reader. The readers were more likely to observe microlobulated morphology in cystic neoplasms than in pseudocysts, with the difference between these lesion types, in terms of prevalence of microlobulated morphology, exhibiting a trend toward-but not reaching-statistical significance (P = .0627). CONCLUSION: Presence of internal dependent debris appears to be a highly specific MR finding for the diagnosis of pancreatic pseudocyst.