RESUMO
BACKGROUND/OBJECTIVES: To analyze the diagnostic performance of three short magnetic resonance imaging (MRI) protocols for the follow-up of pancratic intraductal papillary mucinous neoplasms (IPMN). METHODS: Follow-up MRI examinations of 287 patients with IPMN performed in two centers were retrospectively retrieved. Four MRI protocols were identified as follows: T1-weighted (T1w), T2-weighted (T2w), and MRCP sequences (protocol 1); T1w, T2w, MRCP, and diffusion-weighted (DWI) sequences (protocol 2); T1w, T2w, MRCP, and post-contrast T1w-sequences (protocol 3); and a comprehensive protocol including all previous sequences (protocol 4). Three radiologists with different experience in abdominal imaging expressed their opinion upon the optimal patient's management upon the evaluation of each protocol. Intra-and inter-observer agreement and concordance with the clinical decision expressed by a pancreatic surgeon were calculated with Cohen's kappa test. RESULTS: 223 patients were included (66±10 years; 92 men, 131 women). 143 patients had branch-duct-IPMNs, 25 main-duct-IPMNs and 55 mixed-type-IPMNs. 79 patients underwent surgery, resulting in 52 high-grade dysplasia (HGD) and 27 low-grade dysplasia (LGD). Concordance for the expert reader between protocols 1, 2 and 3 and the actual clinical decision were 0.63, 0.72, and 0.74 respectively (95% CI, 0.53-0.73, 0.63-0.81, and 0.65-0.83). Inter-observer agreement between reader 1 and reader 2, reader 1 and reader 3, and reader 2 and reader 3 were: 0.71, 0.50, and 0.75 for protocol 1 (95% CI, 0.63-0.81, 0.40-0.60, and 0.66-0.84);0.68, 0.54, and 0.84 for protocol 2 (95% CI, 0.59-0.77, 0.44-0.64, and 0.76-0.91); and 0.77, 0.65, and 0.86 for protocol 3 (95% CI, 0.69-0.86, 0.55-0.74, and 0.80-0.93). CONCLUSIONS: Short MRI protocol is suitable for IPMN surveillance.
