Reproducibility of single- and multi-voxel 1H MRS measurements of intramyocellular lipid in overweight and lean subjects under conditions of controlled dietary calorie and fat intake

Abstract

The reproducibility of repeated single-voxel 1H MRS (SV-MRS) and spectroscopic imaging (MRSI) measurements of intramyocellular lipid (IMCL) in the tibialis anterior muscle of five lean and five overweight female Caucasians, during 7 days of controlled dietary fat and calorie intake, was assessed at 1.5 T. Duplicate measures of IMCL relative to total muscle creatine (IMCL/tCr) obtained 3 days apart by both SV-MRS and MRSI correlated well (r = 0.65 and r = 0.95, respectively, P < 0.05). The coefficients of variation for repeated measures of IMCL/tCr by SV-MRS and MRSI were 24.4% and 10.7%, respectively. IMCL/tCr measured by MRSI was higher in overweight subjects than in lean subjects (8.3 +/- 3.8 vs 4.3 +/- 2.4, P < 0.05). Although both methods achieved good reproducibility in measuring IMCL in vivo, MRSI was found to offer greater flexibility and reliability, and higher sensitivity to IMCL differences, whereas SV-MRS was advantageous with respect to shorter scan time and ease of implementation.

Copyright (c) 2007 John Wiley & Sons, Ltd.

Figures

Figure 1

T1-weighted MR images depicting (A) placement of the voxel in the TA muscle for SV-MRS data acquisition, and (B) location of coronal-oblique MRI slice traversing the TA muscle used to prescribe all 1H MRSI scans. The white arrows point to the MRI-visible marker used to help to achieve reproducible voxel or slice positioning in the same subject from scan to scan.

Figure 2

Sample time-domain non-linear least-squares Lorentzian lineshape model fitting of a TA muscle spectrum for deriving IMCL, EMCL and tCr peak areas: (a) measured spectrum; (b) calculated ‘best-fit’ spectrum; (c) individual components of the ‘best-fit’ spectrum; (d) residual of the difference between the measured and calculated ‘best-fit’ spectra.

Figure 3

Representative SV-MRS data of the TA muscle of a lean subject: (A) T1-weighted MR images showing voxels selected in calf muscle regions with (a) no visible intermuscular adipose tissue, (b) minimal visible adipose tissue, and (c) large amount of visible intermuscular adipose tissue. (B) Spectral data from voxels corresponding to those depicted in (A), consisting of resonances for IMCL (1.28 ppm), EMCL (1.50 ppm), and tCr (3.0 ppm). The spectra depicted in (a), (b) and (c) cover the range of spectral patterns observed in this study: (a) IMCL, resonance only, which was uncommon, (b) resolved IMCL and EMCL resonances, which was typical, and (c) a case of severe overlap of the IMCL resonance by that of EMCL.

Figure 4

Representative 1H MRSI data of the same TA muscle as in Fig. 3, showing the spatial distribution of IMCL, EMC, and tCr in a 3 × 3 grid of voxels. Note well-resolved IMCL and EMCL resonances in most voxels, and substantial variations in the intensity of EMCL, whereas those of IMCL and tCr remain relatively constant spatially.

Figure 5

Correlation of first vs second IMCL/tCr values measured by (A) SV-MRS and (B) MRSI for each of 10 subjects.