Clinical Feasibility of Abbreviated Magnetic Resonance With Breath-Hold 3-Dimensional Magnetic Resonance Cholangiopancreatography for Surveillance of Pancreatic Intraductal Papillary Mucinous Neoplasm

Abstract

To determine the clinical feasibility of abbreviated magnetic resonance image (MRI) using breath-hold 3-dimensional magnetic resonance cholangiopancreatography (3D-MRCP) (aMRI-BH) for pancreatic intraductal papillary mucinous neoplasm (IPMN) surveillance. In this retrospective study, 123 patients with 158 pancreatic IPMNs (pathologically proven [n = 73] and typical image feature with ≥2-year stability [n = 85]) who underwent conventional MRI (cMRI) consisting of contrast-enhanced pancreatobiliary MRI with conventional and BH-3D-MRCP were included. Two readers independently evaluated aMRI-BH protocols consisting of heavily T2-weighted, precontrast T1-weighted, and BH-3D-MRCP sequences. The diagnostic performance of aMRI-BH for detecting malignant IPMNs was assessed using the following criteria: category 3, presence of mural nodule 5 mm or bigger and/or main pancreatic duct (MPD) 10 mm or bigger; category 2, more than one of the following: cyst size 30 mm or greater, mural nodule smaller than 5 mm, thickened cyst walls, MPD of 5 to 9 mm, lymphadenopathy, and an abrupt MPD caliber change with distal atrophy; and category 1, none of the above. Categories 2 or 3 were considered positive results of surveillance. Interreader agreement of image features by intraclass correlation and κ statistics were analyzed. The total acquisition times of cMRI and aMRI-BH were 32.7 ± 8 and 5.5 ± 2.1 minutes, respectively (P < 0.01). Among 158 IPMNs, 33 lesions were malignant. The aMRI-BH presented a sensitivity of 100% and a negative predictive value of 100% for evaluating malignant IPMNs in both readers, with substantial interreader agreements (intraclass correlation or к values, range: 0.73-0.93 for cMRI and 0.57-0.94 for aMRI-BH) in significant imaging features based on revised Fukuoka guidelines, except for thickened cyst walls and lymphadenopathy (к values: 0.10 and 1.00 for cMRI and 0.13 and 0.49 for aMRI-BH, respectively). The aMRI-BH provided high sensitivity and negative predictive value to evaluate malignant IPMNs by using predetermined criteria, and aMRI-BH might be a potential tool for pancreatic IPMN surveillance with significantly lower acquisition time.