Measurement of glycine in gray and white matter in the human brain in vivo by 1H MRS at 7.0 T

Abstract

The concentration of glycine (Gly) was measured in gray matter (GM) and white matter (WM) in the human brain using single-voxel localized (1)H MRS at 7 T. A point-resolved spectroscopy sequence with echo time = 150 ms was used for measuring Gly levels in various regions of the frontal and occipital lobes in 11 healthy volunteers and one subject with a glioblastoma. The point-resolved spectroscopy spectra were analyzed with LCModel using basis functions generated from density matrix simulations that included the effects of volume localized radio-frequency and gradient pulses. The fraction of GM and white matter within the voxels was obtained from T(1)-weighted image segmentation. The metabolite concentrations within the voxels, estimated with respect to the GM + WM water concentrations, were fitted to a linear function of fractional GM content. The Gly concentrations in pure GM and white matter were estimated to be 1.1 and 0.1 mM, with 95% confidence intervals 1.0-1.2 and 0.0-0.2, respectively.

FIG 1

Illustration of dependence of LCModel outputs on the baseline spline function parameter, dkntmn. For each of the in-vivo PRESS spectra from medial prefrontal (GM-dominant) and right frontal (WM-dominant) cortices, shown top to bottom are LCModel estimates of concentrations (semi-log scale), CRLBs of Gly and mI, and the correlation coefficients between Gly and mI. The number of signal averages was 64 and 256 for medial prefrontal and right frontal cortices, respectively. TR = 2.5 s, TE = 150 ms, voxel size = 23×23×23 mm³.

FIG 2

In-vivo PRESS spectra from six brain regions of a healthy volunteer are shown together with voxel positioning, LCModel fits, residuals, baseline (red), and spectral components of Gly and mI. Spectra are scaled with respect to the brain GM+WM water signal. LCModel estimates and CRLBs of Gly and mI are shown in brackets. The baseline spline function parameter in LCModel analysis was set to 0.2. Voxels for 3.58 ppm are drawn in yellow and displaced voxels for 2 ppm (due to the effects of the finite bandwidths of volume localization RF pulses) are shown in blue. The ratios of GM and GM+WM contents for the 3.58 ppm voxels are shown for each voxel location. Voxel size was 30×20×15 mm³ in (a), and 23×23×23 mm³ for all other spectra. NSA = 256, 64, 256, 64, 128, and 256 for spectra in (a), (b), (c), (d), (e), and (f), respectively.

FIG 3

(a,b) CRLBs of Gly and mI vs. fractional GM content. (c) Correlation coefficients between Gly and mI vs. fractional GM content. (d,e) Linear regression of Gly and mI levels normalized to the GM+WM water vs. fractional GM content for the 44 voxels positioned in frontal and occipital lobes of the brain. The dashed lines indicate 95% confidence intervals of the linear fits. Circles (brown) and squares (green) represent data from the frontal and occipital lobes, respectively. The baseline spline function parameter dkntmn in LCModel was set to 0.2.

FIG 4

An in-vivo PRESS TE = 150 ms spectrum from a subject with a glioblastoma is shown together with voxel positioning in a T₂w FLAIR image and LCModel fitting results, in a similar fashion to Fig. 2. A vertical dotted line, drawn at 3.62 ppm, indicates an mI signal in the proximity of a large Gly singlet at 3.55 ppm. Data were acquired from a 20×20×20 mm³ voxel, with TR = 2.5 s and NSA = 128.