Test–retest reliability of 3D EPI MR elastography of the pancreas

Abstract

To determine the repeatability of three-dimensional echo planar imaging magnetic resonance (MR) elastography (3D EPI MRE) for assessing pancreatic stiffness in healthy volunteers, patients with chronic pancreatitis, and patients with pancreatic ductal adenocarcinoma (PDAC). 3D EPI MRE was used to assess pancreatic stiffness in five healthy volunteers, and five and three patients with histopathologically confirmed chronic pancreatitis and PDAC, respectively. The study participants underwent two scans, separated by a 1-day interval (referred to as time point 1 and time point 2). Scanning was performed using a 3 T MR machine using a multi-section EPI pulse sequence to assess the mean stiffness of the tumours in PDAC cases, and pancreatic parenchyma in chronic pancreatitis cases and healthy volunteers. A direct-inversion algorithm with 3D post-processing was used to estimate shear stiffness and generate elastograms. Stiffness was measured independently by two analysts. Between the two analysts, the mean stiffness of all 13 pancreases was highly reproducible, with an intraclass correlation coefficient (ICC) of 0.975 (95% confidence interval [CI]: 0.945-0.989) across the two time points (r=0.973, p<0.001). Bland-Altman analysis found a difference in mean stiffness of 0.01kPa (95% agreement limits: -0.53-0.55kPa). Between the two time points, the ICC was 0.973 (95% CI: 0.941-0.988) across the two analysts (r=0.975, p<0.001), and the difference in stiffness by Bland-Altman analysis was 0.05kPa (95% agreement limits: -0.51-0.60kPa). The mean stiffness value of pancreases with PDAC was 2.89±1.20kPa, which was significantly higher than that of normal pancreases (1.11±0.08kPa, p<0.001) and of pancreases with chronic pancreatitis (1.53±0.04kPa, p<0.001). 3D-MRE is a highly reproducible method for assessing pancreatic stiffness.