Associations between iron and mean kurtosis in iron-rich grey matter nuclei in aging

TL;DR

Findings indicate that higher mean kurtosis in iron-rich grey matter structures may be due to either increased tissue complexity or to decreases in signal-to-noise ratios from iron deposition.

Abstract

Elevated kurtosis values have been observed in subcortical grey matter structures of patients with neurodegenerative diseases. Here we tested whether these elevated values are related to iron content, which generate magnetic fields that add to the diffusion encoding gradients. Multi-shell diffusion and multi-echo gradient echo acquisitions were used to derive mean kurtosis and iron measures (R2* and magnetic susceptibility), respectively, in subcortical grey matter nuclei and white matter tracts in a discovery cohort (110 older and 63 younger adults) and replication cohort (72 healthy older adults). Iron-rich grey matter regions exhibited higher mean kurtosis, R2*, and magnetic susceptibility and white matter regions had lower mean kurtosis in the older adult group. In both cohorts, mean kurtosis was significantly correlated with R2* and magnetic susceptibility in iron-rich grey matter nuclei. No association was seen between signal-to-noise ratio and mean kurtosis in any grey matter region, indicating that the increase in mean kurtosis was not due to reduced signal-to-noise. Finally, a phantom experiment found higher mean kurtosis as iron concentrations increased. Our findings indicate that kurtosis is associated with iron-sensitive metrics in iron-rich grey matter structures, suggesting that kurtosis may be sensitive to iron deposits.

Figures (9)

• Figure 1: Denoised images from the$b=1500 \mathrm{~s} / \mathrm{mm}^{2}$ (left column) and $b=3000 \mathrm{~s} / \mathrm{mm}^{2}$ (middle column), and $b=3000 \mathrm{s} / \mathrm{mm}^{2}$ SNR images (right column) in a younger adult (top row) and older adult (bottom row) participant.
• Figure 2: Photographs of the agarose phantom setup. (A) The container with 1 L of water mixed with$2.5 \%$ agarose. Four vials contained a mixture of water with $2.5 \%$ agarose and different concentrations of ferric citrate ( $0.03 \mathrm{mMol}, 0.06 \mathrm{mMol}, 0.09 \mathrm{mMol}$, and 0.12 mMol ) and one vial contained deionized water. (B) An image of the phantom inside a 32channel head coil.
• Figure 3: Group average images for mean diffusivity (first column) mean kurtosis (second column),$\mathrm{R}_{2}{ }^{*}$ (third column), and susceptibility (fourth column) in younger (top row) and older (bottom row) participants. These images were created by transforming each participant's mean diffusivity, mean kurtosis, $\mathrm{R}_{2}{ }^{*}$ map, or susceptibility map to Montreal Neurological Institute (MNI) common space and averaging within each group
• Figure 4: Group comparisons of mean kurtosis ( MK ; shown in A ) and$\mathrm{R}_{2}{ }^{*}$ (shown in B) for all grey matter ROIs considered in this analysis. Significant increases in MK and $\mathrm{R}_{2}{ }^{*}$ were observed in the pallidum, putamen, dentate nucleus, caudate nucleus, and hippocampus of the older adult group relative to the younger adult group. No difference in MK or $\mathrm{R}_{2}$ * was seen in the thalamus.
• Figure 5: Group comparisons of mean kurtosis in the superior longitudinal fasciculus, forceps major, and forceps minor are shown in the top row (A-C). Within older adults, associations between age and MK in these white matter tracts are shown in the bottom row (D-F). Significant correlations were seen between MK and age in each white matter tract$\left(P s<10^{-3}\right)$.