Evaluating g-ratio weighted changes in the corpus callosum as a function of age and sex

Abstract

Recent years have seen a growing interest in relating MRI measurements to the structural-biophysical properties of white matter fibers. The fiber g-ratio, defined as the ratio between the inner and outer radii of the axon myelin sheath, is an important structural property of white matter, affecting signal conduction. Recently proposed modeling methods that use a combination of quantitative-MRI signals, enable a measurement of the fiber g-ratio in vivo. Here we use an MRI-based g-ratio estimation to observe the variance of the g-ratio within the corpus callosum, and evaluate sex and age related differences. To estimate the g-ratio we used a model (Stikov et al., 2011; Duval et al., 2017) based on two different WM microstructure parameters: the relative amounts of myelin (myelin volume fraction, MVF) and fibers (fiber volume fraction, FVF) in a voxel. We derived the FVF from the fractional anisotropy (FA), and estimated the MVF by using the lipid and macromolecular tissue volume (MTV), calculated from the proton density (Mezer et al., 2013). In comparison to other methods of estimating the MVF, MTV represents a stable parameter with a straightforward route of acquisition. To establish our model, we first compared histological MVF measurements (West et al., 2016) with the MRI derived MTV. We then implemented our model on a large database of 92 subjects (44 males), aged 7 to 81, in order to evaluate age and sex related changes within the corpus callosum. Our results show that the MTV provides a good estimation of MVF for calculating g-ratio, and produced values from the corpus callosum that correspond to those found in animals ex vivo and are close to the theoretical optimum, as well as to published in vivo data. Our results demonstrate that the MTV derived g-ratio provides a simple and reliable in vivo g-ratio-weighted (GR*) measurement in humans. In agreement with theoretical predictions, and unlike other tissue parameters measured with MRI, the g-ratio estimations were found to be relatively stable with age, and we found no support for a significant sexual dimorphism with age.

Keywords: Age; Corpus callosum; Sex; g-ratio.

Figure 1. GR* model validation

a. MTV_{ex vivo}, the MRI-calculated non-water content, measured with T2 spectrum, as a function of the histological myelin volume fraction, MVF_hist (coefficient of determination, R² = 0.73). Each data point represents a different mouse model of abnormal myelination, and an average over several animals and three regions along the corpus callosum. Data taken from West et al. (2016). b. The reproducibility of the human MRI-derived GR* measurement, as the correlation between two separate scans (R² =0.51). The plot represents the mean GR* (± std) of eight corpus callosum sub-regions from four subjects. The Bland-Altman plots of the same data are presented in supplementary figure 2.

Figure 2. FA stability across b-values

a 2D histogram of all of the FA values measured with two different diffusion weightings. The values presented here are from the corpus callosum of all subjects (N=92). The agreement between FA of different diffusion weightings, quantified with the coefficient of determination, is high (R² = 74).

Figure 3. GR* model vs. literature values

a. histogram of GR* values in the corpus callosum of all subjects, $\widetilde{G{R}^{\ast }}$. b. Mean GR* for each callosal sub-region in the test subjects, projected on the R1 map and callosal segmentation from a single subject. The value of GR* changes as a function of callosal sub-region (p < 0.001). c. Comparison of GR* and histological g-ratio (from Stikov et al., 2015) in the splenium, mid-body and genu regions, showing the anterior-posterior trend in these different areas. The image below the chart shows the fibers crossing through the sub-regions of the corpus callosum.

Figure 4. GR* change with age

Each subplot shows results for a different callosal sub-region, (colors correspond to the arbitrary segmentation in the final image in the lower right panel). For each sub-region, the mean GR* of the subjects is plotted as a function of age, as well as a regression line. In two of the eight sub-regions; the Motor Fiber crossing area and the Anterior Frontal crossing area, we found a significant correlation of GR* with age. Each subplot notes the correlation coefficient (ρ) and the corresponding p-value (p).

Figure 5. GR* changes with age and sex

Each subplot shows the results for a different callosal sub-region (color coded). For each sub-region, the mean GR* of males and females of ages 8 to 50, is plotted as a function of age. In two of the eight sub-regions, the Motor and Anterior-Frontal regions, we found a significant correlation of GR* with age. We found no significant differences (corrected for multiple comparisons) between the development of GR* of males versus females, in any of the sub-regions.