1927
Spatially resolved measurement of lipid droplet size in white adipose tissue with high b-value stimulated echo-prepared diffusion-weighted 2D single shot TSE1Department of Diagnostic and Interventional Radiology, Technical University Munich, Munich, Germany, 2Else Kröner Fresenius Center for Nutritional Medicine, Technical University Munich, Munich, Germany, 3Chair for Food and Bioprocess Engineering, Technical University Munich, Freising, Germany
Synopsis
Despite its strong relevance in metabolic dysfunction, non-invasive measurement of fat microstructure remains an unmet need. In white adipose tissue, enlarged adipocyte size is linked to the obese phenotype. DW-MRS has been previously applied to probe diffusion restriction effects of intramyocellular lipids or brown adipocytes using preclinical systems. However, probing diffusion restriction in large lipid droplets remains a major challenge and DW-MRS can only measure spatially averaged effects. This work proposes a method that probes lipid droplet sizes with high b-value stimulated-echo prepared DW 2D single shot TSE, validates the methods in water-fat phantoms and applies it in the gluteal fat depot in vivo.
Purpose
The measurement of lipid droplet size is important in the study of adipose tissue (AT) and ectopic lipids in both health and metabolic dysfunction across organs and tissues. In fat depots containing brown AT, brown adipocytes consist of much smaller lipid droplets than white adipocytes1. In white AT, enlarged adipocyte size is linked to the obese phenotype2. However, the assessment of AT droplet size currently requires a highly invasive biopsy procedure. Diffusion-weighted (DW) magnetic resonance (MR) is a powerful tool for the non-invasive assessment of tissue microstructure. The reduction of the apparent diffusion coefficient with increasing diffusion times due to diffusion restriction effects has been previously applied to extract cell size in water-containing tissues3. Measuring the diffusion properties of lipids has proven to be challenging because fat has a diffusion coefficient approximately two orders of magnitude lower than water4,5. A low lipid diffusion coefficient increases the required diffusion weighting, inducing technical challenges related to eddy currents6 and increased sensitivity to macroscopic motion7. Single-voxel DW-MR spectroscopy (DW-MRS) has been successfully applied to investigate lipid diffusion properties4 and was recently applied to probe lipid diffusion restriction effects8. However, DW-MRS can only measure spatially averaged properties and therefore lipid DWI is highly desirable in spatially heterogeneous fatty tissue. A high b-value stimulated-echo prepared DW 2D single shot TSE (STE-DW TSE) sequence was recently proposed to probe spatially resolved lipid diffusion9. This work proposes a novel method that probes lipid droplet sizes with the proposed STE-DW TSE, validates the methods in phantoms and applies it in vivo in the gluteal fat.Methods
Background:
The signal decay assuming restricted diffusion and spherical boundaries can be described by10:
$$$ ln\left(\frac{S\left(\Delta,\delta,G\right)}{S_{0}\left(\Delta\right)}\right)=-2\gamma^{2}G^{2}\sum_{m=1}^{\infty}\left[\alpha_{m}^{2}\left(\alpha_{m}^{2}\left(\frac{d}{2}\right)^{2}-2\right)\right]^{-1}\\*\left(\frac{2\delta}{\alpha_{m}^{2}D}-\frac{2+exp\left(-\alpha_{m}^{2}D\left(\Delta-\delta\right)\right)-2exp\left(-\alpha_{m}^{2}D\delta\right)-2exp\left(-\alpha_{m}^{2}D\Delta\right)+exp\left(-\alpha_{m}^{2}D\left(\Delta+\delta\right)\right)}{\left(\alpha_{m}^{2}D\right)^{2}}\right)\quad\left(1\right)$$$
where S: DW signal, S0: Non-DW signal, ∆: diffusion time, δ: diffusion gradient length, G: gradient strength, d: restriction barrier diameter, D: free diffusion constant and αm: roots obtained by a separate differential equation. Equation 1 can be employed to extract d:
$$$\underset{D,d,T_{1},\rho}{arg\min}\left\lvert\left\lvert S\left(\Delta,\delta,G,T_{1}\right)*\rho*exp\left(-\frac{TM}{T_{1}}\right)-S_{exp}\right\rvert\right\rvert\quad\left(2\right)$$$
Pulse sequence:
A STE-DW preparation consisting of four 90° RF pulses and mono-polar diffusion gradients, followed by a single-shot 2D TSE readout, was used (Figure 1). To mitigate artifacts, vibration compensating gradients were utilized.
Phantom:
Water-fat phantoms closely resembling AT were produced (content: 800ml oil, 200ml water, 4ml Tween80, 1g of sodium benzoate). Emulsification was carried out with a colloid mill at 5000/6000/9000/12000 revolutions per minute to vary oil droplet sizes. The phantoms were scanned on a 3T system (Ingenia Elition, Philips, Best) using an 8-channel wrist coil with the following parameters: FOV: (90x90)mm2, voxel size: (1.4x2.5x10)mm3, TR/TE/TEPrep: 2000/23/61ms, 5 averages, 50 dynamics with b-values: 5,000s/mm2 to 50,000s/mm2 in 5,000s/mm2 steps and TMPrep: 200/250/300/350/400ms, scan time: 8:30min. To minimize vibration effects, a wooden support table was utilized. For validation, the particle size was measured by dynamic light scattering (Mastersizer 2000, Malvern Instruments, Worcestershire).
In vivo:
The gluteal fat was scanned three times without repositioning in 5 volunteers (male/female: 1/4, mean age: 27.6±1.8) with a 12-channel posterior and 16-channel anterior coil. The sequence parameters were: FOV(unilateral): (200x110)mm2, voxel size: 3x3x3mm3, TR/TE: 2400/23ms, 4 averages, respiratory triggering (delay: 600ms), average scan time: 13min. The remaining parameters matched the phantom scans.
Post-Processing:
The magnitude signal of the DW images was fitted to equation 2 voxel-by-voxel. Background noise was subtracted11,12 and averages that deviated more than one standard deviation from the mean value were excluded. In two volunteers (#2,#4), the first TM was excluded due to artifacts.
Results
Figure 2 shows exemplary mean signal decay curves with corresponding fitting and diameter maps for the phantoms. A trend towards smaller lipid droplet sizes was observed for increasing stirring frequency. Figure 3 shows the comparison between droplet sizes obtained by MR and laser deflection. High correlation coefficients (R2/p: 0.99/0.0027) were observed. In Figure 4, exemplary in vivo diameter maps are shown. The three repeated scans were in close agreement and a relatively homogenous voxel-wise coefficient of variation (COV) below 20% was found over the fat region. The posterior part of the gluteal fat was segmented to extract mean diameter and COV for all subjects (Figure 5). The mean lipid droplet size was 71.3±6.1µm with a mean COV of 14.5±2.4%Discussion and Conclusion
To the best of the authors’ knowledge, this is the first time that droplet sizes are measured in large adipocytes using DWI in vivo. The phantom validation experiments show a very good agreement with the reference measurement. The obtained in vivo lipid droplet sizes agree closely with recently reported white adipocytes sizes13. An MR-based white adipocyte size estimate could potentially serve as a biomarker in the metabolic syndrome and would be of high interest in longitudinal lifestyle intervention studies.Acknowledgements
The present work was supported by the European Research Council (grant agreement No 677661, ProFatMRI) and Philips Healthcare. This work reflects only the authors view and the EU is not responsible for any use that may be made of the information it contains.References
1. Cousin B, Cinti S, Morroni M, Raimbault S, Ricquier D, Penicaud L, Casteilla L. Occurrence of brown adipocytes in rat white adipose tissue: molecular and morphological characterization. J Cell Sci 1992;103 ( Pt 4):931-942.
2. Salans LB, Cushman SW, Weismann RE. Studies of human adipose tissue. Adipose cell size and number in nonobese and obese patients. J Clin Invest 1973;52(4):929-941.
3. Assaf Y, Blumenfeld-Katzir T, Yovel Y, Basser PJ. AxCaliber: a method for measuring axon diameter distribution from diffusion MRI. Magn Reson Med 2008;59(6):1347-1354.
4. Lehnert A, Machann J, Helms G, Claussen CD, Schick F. Diffusion characteristics of large molecules assessed by proton MRS on a whole-body MR system. Magn Reson Imaging 2004;22(1):39-46.
5. Steidle G, Eibofner F, Schick F. Quantitative Diffusion Imaging of Adipose Tissue in the Human Lower Leg at 1.5 T. Magnetic Resonance in Medicine 2011;65(4):1119-1125.
6. Jezzard P, Barnett AS, Pierpaoli C. Characterization of and correction for eddy current artifacts in echo planar diffusion imaging. Magn Reson Med 1998;39(5):801-812.
7. Anderson. Analysis and correction of motion artifacts in diffusion weighted imaging. Magn Reson Med 1994;32(3):379-387.
8. Weidlich D. HA, Ruschke S., Franz D., Steiger K., Skurk T., Hauner H., Rummeny E.J., Karampinos, D. C. Probing bone marrow adipocyte cell size in vivo at a clinical 3 T scanner using high b-value DW-MRS at long diffusion times. 2017; USA, Hawaii, Honolulu. p 1227.
9. Weidlich D. RS, Cervantes B., Hock A., Karampinos, D. C. ADC quantification of lipids with high b-value stimulated echo-prepared diffusion- weighted 2D single shot TSE. 2018; Paris, France. p 2511.
10. Murday JS, Cotts RM. Self-Diffusion Coeffcient of Liquid Lithium. Journal of Chemical Physics 1968;48(11):4938-&.
11. Miller AJ, Joseph PM. The Use of Power Images to Perform Quantitative-Analysis on Low Snr Mr-Images. Magnetic Resonance Imaging 1993;11(7):1051-1056.
12. Raya JG, Dietrich O, Horng A, Weber J, Reiser MF, Glaser C. T2 measurement in articular cartilage: impact of the fitting method on accuracy and precision at low SNR. Magn Reson Med 2010;63(1):181-193.
13. Skurk T, Alberti-Huber C, Herder C, Hauner H. Relationship between adipocyte size and adipokine expression and secretion. Journal of Clinical Endocrinology & Metabolism 2007;92(3):1023-1033.
Figures
Figure 1:
Sequence diagram of the proposed STE-DW single shot 2D TSE. The diffusion weighted STE-DW magnetization preparation consists of four composite 90° RF pulses, mono-polar diffusion sensitizing gradients and a spoiler gradient during the mixing time. To eliminate motion-induced phase errors an additional pair of de-/rephasing gradients (indicated in red) denoted as magnitude stabilizers are introduced before the last tip-up pulse. Magnitude stabilizers are performed immediately before and after every spin echo formation. Vibration compensating gradients matching the diffusion gradient properties were placed before the diffusion preparation to mitigate vibration artifacts.
Figure 2:
Mean signal decay curve in two different phantoms with corresponding fitting of equation 2 (upper row). All five averages for each b-value and mixing time are treated as independent measurement points. In the 5,000 rpm phantom compared to the 12,000 rpm phantom a stronger diffusion decay is observed, indicating larger diffusion restriction barriers. In the bottom row the phantom diameters obtained by a voxel-by-voxel fitting are shown. On average decreasing droplet diameters are observed at increasing emulsification rotation frequencies.
Figure 3:
The volume particle size distributions in water-fat phantoms measured with laser deflection (a). Slower rotation frequencies in the emulsification process lead to larger oil droplets in the water fat phantoms. (b) shows the mean diameter obtained by DW-TSE in comparison with the laser deflection validation measurement. (c) shows the correlation analysis of the two measurements. The R2 coefficients are in good agreement between DW-MRS and laser deflection (R2 = 0.99, p=0.0027). The slope was 1.02 and the offset -0.53 µm.
Figure 4:
Anatomical survey (a) shows approximate height of the acquired 2D slice and PDFF map (e) shows the acquired 2D slice in the gluteal fat unilateral. (b), (c) and (d) show the diameter maps in one volunteer at three repetitive scans. (f) shows the mean diameter, (g) the standard deviation of the mean diameter and (h) a coefficient of variation (COV) map. The COV stays below 20% for large parts of the gluteal fat depot. In (c) the posterior gluteal fat ROI defined in all subjects and used for the subsequent analysis is indicated.
Figure 5:
Mean lipid droplet diameters and mean coefficient of variation (COV) obtained in the posterior gluteal fat depot by DW-TSE measurements. The overall mean diameter obtained from the volunteer study was 71.3 ± 6.1 µm whereas the mean COV stayed below 20% for all measured subjects. The BMI is also shown for each subject.