1940
Assessing variability in MRI-based quantitative measurements of body fat in patients with NASH1Department of Radiology, Duke University Medical Center, Durham, NC, United States, 2Pattern Recognition Lab, Department of Computer Science, Friedrich‐Alexander‐Universität Erlangen‐Nürnberg, Erlangen, Germany, 3Department of Biostatistics and Bioinformatics, Duke University School of Medicine, Durham, NC, United States, 4Department of Biostatistics and Bioinformatics, Duke University Medical Center, Durham, NC, United States, 5MR Applications Predevelopment, Siemens Healthcare GmbH, Erlangen, Germany, 6Center for Advanced Magnetic Resonance Imaging, Duke University Medical Center, Durham, NC, United States, 7Department of Medicine, Gastroenterology, Duke University Medical Center, Durham, NC, United States
Synopsis
NASH is a common cause of chronic liver disease, and is closely associated with other metabolic derangements; quantifying types of fat (saturated, mono-unsaturated, poly-unsaturated) in the body may provide insights into metabolism. We assessed a chemical-shift encoded MRI technique for quantifying types of fat in various depots. Inter-reader repeatability in the subcutaneous tissues and retroperitoneum was high (mean 0.86-0.90), and similar to PDFF (mean 0.89). Correlation between fat types was moderate to strong between non-liver fat depots, and weaker and inconsistent between liver and non-liver fat depots.
Introduction
Chemical shift-encoded MRI can estimate liver fat content, which is relevant to non-alcoholic steatohepatitis (NASH), and can also be used to assess different types of fat (saturated, mono-unsaturated, poly-unsaturated) deposited in the liver and other fat depots. The types of fat deposited in these tissues have been linked with histopathological findings of NASH1,2. We assessed an MRI-based technique designed to provide quantitative measures of the proportions of different types of fat in various tissues, expressed as a percentage of the total fat 3. Our purpose was to assess sources of variability in fat composition measurements, including readers, spatial location, and anatomic location of the fat depots.Methods
This was an IRB approved, HIPAA compliant retrospective analysis of MRI data collected prospectively in a clinical trial investigating drug therapy for NASH. A total of 33 patients with suspected NASH were included, with 52 total scans. 19 patients had paired pre- and post-treatment MRI with the pulse sequence detailed below; the remaining 14 patients only had either pre- or post-treatment MRI with the required sequence. The study population included 19 women/14 men, age 46±13 years.
Imaging was performed on 3T MRI systems (MAGNETOM Trio or Skyra, Siemens Healthcare, Erlangen, Germany). The acquisition was a bipolar multi-echo 2D GRE with 12 echoes acquired, first TE and TE spacing=1.19 msec, flip angle = 10°, bandwidth = 1955 Hz/px, TR = 175 msec, matrix = 128 x 116. Triglyceride saturation state parameters were then calculated offline using prototype post-processing by considering confounding factors (B0 off-resonances, R2*, eddy-current phase) from low-rank denoised multi-echo images4. This approach relies on an analytically derived formulation for the eddy-current-induced phase discrepancies between even and odd echoes.
Three radiologists independently performed measurements on each of PDFF, saturated, mono-unsaturated, and poly-unsaturated fat maps for each patient at 14 different locations including: subcutaneous tissues (8 locations permuting anterior/posterior, upper/lower, and right/left); liver (3 locations); retroperitoneum (2 locations right/left); and mesentery (single location).
Inter-reader repeatability of measures was calculated using Cronbach’s Alpha (standardized). Variability due to differences in spatial location in the x-, y-, and z-directions was assessed using pairs of ROIs in the subcutaneous regions with the ICCs computed using mixed-effects ANOVA and Cronbach's Alpha. Finally, variability based on fat depot was assessed between subcutaneous, retroperitoneal, mesenteric, and liver fat using Spearman’s correlation.
Results
Inter-reader repeatability in the subcutaneous tissues was: saturated fat mean 0.86 (range 0.74-0.94); mono-unsaturated fat mean 0.90 (range 0.86-0.94); poly-unsaturated fat mean 0.89 (range 0.87-0.93); these were comparable to inter-reader repeatability for PDFF (mean 0.89, range 0.88-0.91). Repeatability was similar in the retroperitoneum, but lower in the mesentery (0.59-0.75 for the types of fat, 0.43 for PDFF).
Significant differences in measured values from right to left were encountered for several values for the Trio system. Saturated fat was most significantly affected (p=0.006 to p=0.19) with an average difference of 7.3% between right and left locations. The remaining values for the Trio system and values for the Skyra system were less affected.
The proportions of different types of fat were moderately to strongly correlated (rho=0.35-0.84, p<0.02 to p<0.001) between subcutaneous, retroperitoneal, and mesenteric fat; correlations between fat types in the liver and other locations were much weaker and less significant (rho=-0.02-0.40, p=0.94 to p<0.005).
Discussion
Inter-reader repeatability for this technique was relatively high, but there remain challenges with value gradients in the frequency-encoding direction. Strong correlations between non-liver fat depot values suggest that it may not be necessary to measure all of these separately.Conclusion
Chemical-shift-encoded MRI offers insights into metabolism related to body fat beyond simple liver fat quantification. Some technical challenges remain, but high repeatability with the current technique suggests that this could become a reliable biomarker.Acknowledgements
No acknowledgement found.References
1. Leporq B, Lambert SA, Ronot M, et al. Simultaneous MR quantification of hepatic fat content, fatty acid composition, transverse relaxation time, and magnetic susceptibility for the diagnosis of non-alcoholic steatohepatitis. NMR Biomed. 2017 Oct;30(10).
2. Flintham R, Eddowes P, Semple S, et al. Non-invasive quantification and characterization of liver fat in non-alcoholic fatty liver disease (NAFLD) using automated analysis of MRS correlated with histology. Proceedings of the International Society of Magnetic Resonance in Medicine, May 2016.
3. Schneider M, Janas G, Lugauer F, et al. Accurate fatty acid composition estimation of adipose tissue in the abdomen based on bipolar multi-echo MRI. Magnetic Resonance in Medicine. 2018. doi: 10.1002/mrm.27557.
4. Lugauer F, Nickel D, Wetzl J, et al. Robust Spectral Denoising for Water-Fat Separation in Magnetic Resonance Imaging. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. Springer; 2015. p. 667-674.
Figures
