Predicting post-operative pancreatic fistulae using preoperative pancreatic imaging: a systematic review

Post-operative pancreatic fistulae (POPF) remain a major contributor to morbidity and mortality following pancreatic resection. Evidence for preoperative prediction of POPF based on cross-sectional imaging has not been systemically reviewed. This review aimed to determine whether preoperative imaging modalities can accurately predict the development of POPF.

A systematic review of major reference databases was undertaken, according to PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, up to May 2018.

There were 18 studies (2150 patients), seven used magnetic resonance imaging (MRI), five used computed tomography (CT) scans, four used transabdominal ultrasonography and one study each used MRI and CT and endoscopic ultrasonography elastography. All were retrospective, single-centre studies. Intensity of the pancreas signal relative to the spleen, liver or muscle was commonly used. Other studies compared signal intensity between unenhanced and post-contrast-enhanced pancreas, apparent diffusion coefficient values comparing normal parenchyma to fibrosis, perfusion fraction (f) of intravoxel incoherent motion diffusion-weighted imaging, or utilized a muscle-normalized signal intensity curve with signal intensity ratio or directly assessed pancreatic volume and duct width. Shear wave velocity measurement on transabdominal ultrasonography may reflect pancreas tissue fibrosis or stiffness and predict POPF. Most parameters used to predict the development of POPF were based on identifying imaging features of a fatty or fibrotic pancreas and main pancreatic duct diameter.

A number of different and highly promising parameters have been used for preoperative prediction of POPF using ultrasound, MRI, CT or both. Large multicentre prospective studies are needed to determine which parameters most accurately predict POPF, using standardized definitions and methodology.