Global brain metabolic quantification with whole-head proton MRS at 3 T

Abstract

Total N-acetyl-aspartate + N-acetyl-aspartate-glutamate (NAA), total creatine (Cr) and total choline (Cho) proton MRS (¹ H-MRS) signals are often used as surrogate markers in diffuse neurological pathologies, but spatial coverage of this methodology is limited to 1%-65% of the brain. Here we wish to demonstrate that non-localized, whole-head (WH) ¹ H-MRS captures just the brain's contribution to the Cho and Cr signals, ignoring all other compartments. Towards this end, 27 young healthy adults (18 men, 9 women), 29.9 ± 8.5 years old, were recruited and underwent T₁ -weighted MRI for tissue segmentation, non-localizing, approximately 3 min WH ¹ H-MRS (T_E /T_R /T_I = 5/10/940 ms) and 30 min ¹ H-MR spectroscopic imaging (MRSI) (T_E /T_R = 35/2100 ms) in a 360 cm³ volume of interest (VOI) at the brain's center. The VOI absolute NAA, Cr and Cho concentrations, 7.7 ± 0.5, 5.5 ± 0.4 and 1.3 ± 0.2 mM, were all within 10% of the WH: 8.6 ± 1.1, 6.0 ± 1.0 and 1.3 ± 0.2 mM. The mean NAA/Cr and NAA/Cho ratios in the WH were only slightly higher than the "brain-only" VOI: 1.5 versus 1.4 (7%) and 6.6 versus 5.9 (11%); Cho/Cr were not different. The brain/WH volume ratio was 0.31 ± 0.03 (brain ≈ 30% of WH volume). Air-tissue susceptibility-driven local magnetic field changes going from the brain outwards showed sharp gradients of more than 100 Hz/cm (1 ppm/cm), explaining the skull's Cr and Cho signal losses through resonance shifts, line broadening and destructive interference. The similarity of non-localized WH and localized VOI NAA, Cr and Cho concentrations and their ratios suggests that their signals originate predominantly from the brain. Therefore, the fast, comprehensive WH-¹ H-MRS method may facilitate quantification of these metabolites, which are common surrogate markers in neurological disorders.

Keywords: N-acetyl-aspartate; choline; creatine; diffuse brain disorders; normal brain; proton MRS; whole-head MRS.

Fig. 1

Top: WH ¹H-MRS from a 2 L phantom (a) and three subjects (b – d, numbers 6, 10 and 16 in Table 1), from the same session as a’ – d’, below (thin black line). All on the same frequency and intensity scales, overlaid with their VeSPA analysis (fit – thick gray line) and residual (raw – fit) from 0 – 4 ppm. Note (i) the vanishing residual, reflecting VeSPA fit’s quality; (ii) excellent lipid suppression performance; and (iii) the qualitative similarity between the WH-¹H-MRS brain+head signal, a – d, and localized MRS (brain only) a’ – d’. Bottom, a’ – d’: Raw data (sum of 480 spectra in the 8×10×4.5 cm³ VOI, e.g., Fig. 3d) overlaid with SITools-FITT baseline + model functions from the phantom (a’) and three above subjects (b’ – d’), all on the same frequency and intensity scales. Beneath the spectra is the (raw – fit) residual. Note (i) the vanishing residual, reflecting the fit quality; and (ii) that the WH sequence a 13̅31̅ water and lipids suppressing readout pulse includes the NAA, Cho and Cr in its passband, but excludes the mI at 3.56 ppm [see reference (11); and that despite a linear phase roll across that passband that puts the Cr and Cho peaks almost at an anti-phase, the VeSPA spectral fitting tool and customized basis functions handle this lineshape appropriately, as seen in the figure and quantified at a better than 7% intra-subject inter-scan reproducibility (69)].

Fig. 2

Top, left: Sagittal (a) and Axial (b) T1-weighted MRI from a male subject superimposed with the 8×10×4.5 cm³ (LR×AP×IS) VOI and 16×16 cm² axial CSI FOV (solid and dashed lines). The white arrow on a indicates the level of b. Right, c: Real part of the 8×10 axial (LR×AP) ¹H spectra matrix from the VOI slice shown on a and marked with the solid white arrow. All spectra are on a common frequency (ppm) and intensity scale. Note the good SNR and excellent spectral resolution (6±2 Hz linewidth) from these high spatial resolution (0.75 cm³) voxels. Bottom, left, d: the sum of all 480 spectra in the VOI (pre-aligned to the NAA peak) (thin black line) overlaid with the spectral fitting (thick gray line). Note the spectral resolution (7±1 Hz linewidth), exceedingly high SNR from this 360 cm³ VOI, and fidelity of the fit.

Fig. 3

Box plots showing the first, second (median) and third quartiles, ±95^th percentiles (whiskers), outliers (*) and means (+) of the WH and VOI absolute NAA, Cr and Cho concentrations distribution in N=27 and N=23 subjects. Note the (i) median and mean proximity, suggesting a normal distribution; (ii) that although the WH NAA and Cr is significantly higher than in the VOI (marked by arrows) they are nevertheless of similar magnitude: 12%, 9% and 3% (Cho) differences between their means in the WH and VOI.

Fig. 4

Box plots of the Cho/Cr, NAA/Cr and NAA/Cho ratios distributions obtained with WH ¹H-MRS (gray), N=27 and with 3D ¹H-MRSI from a VOI inside the brain (N=23). Note (i) similar WH and VOI Cho/Cr; (ii) median and mean proximity, suggesting normal distributions; (iii) Although WH and VOI NAA/Cr and NAA/Cho are technically “different” (marked by arrows), these differences are minimal, 7% and 11%, not the expected ~300%. The similarities suggest that WH ¹H-MRS comprises overwhelmingly brain, not other head tissue metabolites’ signals.

Fig. 5

Top, a: Sagittal B₀ shift (ΔB₀) maps from a 29-year-old female (number 9 in Table 1) at (2 mm)³ spatial resolution. The b – o scale (gradations at 1 cm steps) indicates the levels of the axial ΔB₀ maps: b – o, below Bottom, b – o: Axial ΔB₀ maps corresponding to the levels of the scale on a, above, extending from the vertex to the foernum magnum. Note (i) the small ΔB₀s reflecting relatively uniform field, with only small, ≤10 Hz, variations within the brain – cerebrum and cerebellum; and (ii) the dramatically larger, ›100 Hz/cm variations at the brain-skull-air transitions and near internal air-tissue interfaces, e.g., ear canals, air-filled sinuses and face (slices m – o). These are likely responsible for effectively suppressing the Cho and Cr signals from these transition regions.

Fig. 6

Histogram of the WH 3.65×10⁵ pixels B₀ shifts (ΔB₀) distribution of the subject in Fig. 5 (number 9 in Table 1) at 10 Hz/bin resolution. Note (i) that the dominant first bin, comprises 145×10⁴ pixels (1160 cm³) with the most homogeneous, ΔB₀≤±10 Hz, (corresponds to local susceptibility ≤0.08 ppm/cm) equals ~90% of her brain volume, therefore, likely originate in the gray regions in Fig. 5, i.e., just the brain. Therefore, the rest of the pixels that sustain greater local ΔB₀s must comprise compartments extraneous to the brain (see Fig. 5) and their metabolites signal is lost to susceptibility broadening and shifts.