Synopsis

Keywords: Neurodegeneration, Aging

In a large cohort of healthy participants (N=349, age range 18-90 years), we characterized intracortical myelin lifespan trajectories using two commonly used myelin proxies, magnetization transfer ratio (MTR) and T1w/T2w, combined with a surface-based image processing method. We showed that both measures exhibit an inverted-U shape trajectory for most cortical regions, with T1w/T2w values peaking later than MTR. A smoothed spline fitting of the trajectories allowed further delineation of onset decline age, a potential aging milestone. Frontal and temporal lobe regions showed consistently later peak ages than parietal and sensory motor regions across MRI metrics and curve fitting methods.

Purpose

Myelination in the human brain has been shown to follow an inverted-U shape trajectory across the lifespan [1-3]. Intracortical myelin, which associates with higher order functions, has a more protracted development and substantial regional variations [4]. With the advances in in vivo myelin mapping using MRI, intracortical myelination during development and aging can be better understood by studies spanning sufficient age ranges [5] and utilizing multiple imaging methods [6]. The complexity of the cortex also calls for novel image processing techniques: distinct cytoarchitectural layers may differ in iron content [7] and presence of crossing myelinated fibers [8]. These variations may bias traditional MRI volumetric intracortical myelin approximation methods.

To address current gaps, we aimed to characterize and compare trajectories of intracortical myelin from 18 to 90 years of age using magnetization transfer ratio (MTR) [9] and T1w/T2w [10] combined with a surface-based image processing method. We hypothesized that both measures would show inverted-U shape trajectories but with specific curve features given that each of these measures may be influenced by factors other than myelin.

Methods

We utilized data from healthy participants (ages 18.5-88.9) collected by the Cambridge Centre for Ageing and Neuroscience (Cam-CAN) project. The recruitment and MRI acquisition has been thoroughly described in previous papers [11, 12].

Here, we used Cam-CAN high resolution T1w, T2w, and magnetization transfer (MT) images. The T1w and T2w structural images were acquired with 1mm isotropic voxels. Other main acquisition parameters included TR=2250ms, TE=2.99ms, TI=900ms, FA=9º for T1w MPRAGE, and TR=2800ms, TE=408ms for T2w SPACE. MT images were acquired using prepared SPGR sequences with TR=30ms, TE=5ms, FA=12º, and 1.5x1.5x1.6mm3 voxels. Off resonance saturation was achieved by applying a Gaussian pulse with FA=500º, pulse duration=9.984ms, and offset frequency=1950Hz.

We processed the structural images using the Human Connectome Project (HCP) preprocessing pipeline [13], which includes bias correction, surface reconstruction, and T1w/T2w calculation. The MT images were corrected for Gibbs ringing before being registered to the structural space [14, 15]. MTR maps were calculated as (M0-Ms)/M0, where Ms and M0 are signal intensities with and without saturation pulse. After image quality control, 349 participants were included in the analysis (Figure 1).

For intracortical myelin estimation, we used MTR and T1w/T2w value at 70% cortical thickness depth to minimize partial volume averaging from non-grey matter tissue [5]. FreeSurfer tools [16] were employed to expand surfaces (Figure 1B) and extract mean values from each region of interest (ROI) based on the HCP multimodal parcellation [17].

To characterize intracortical myelin trajectory, we first tested quadratic age effects with sex and linear age included as covariates in the regression. Multiple comparisons were corrected using the Benjamini-Hochberg method. For regions showing a significant quadratic age effect, we used a smoothing spline fitting (Matlab csaps) to further identify curve features. Derivatives were calculated for fitting curves to locate ages of peak myelination for quadratic fitting and onset age of decline for spline fitting (last peak in curve). All analyses were performed in Matlab.

Results

Intracortical myelin estimated by MTR and T1w/T2w showed similar distribution pattern across regions, with primary areas such as the sensory motor region showing higher values compared to association areas (Figure 2A-2B). While differences were observed in the occipital lobe regions, significant correlations were found between measures both across regions and across subjects (Figure 2C-2D).

Inverted-U shape trajectories were observed in extended areas (Figure 3A-3B). Peak ages of the quadratic curve from the two measures showed a significant correlation. However, MTR values generally peaked much earlier than the T1w/T2w ratio (Figure 3C-3D).

Spline fitting of the trajectories revealed unique curve features. For MTR particularly, a relatively flat myelination period appeared during middle age (Figure 4) and the onset age of decline was delayed compared to the previously identified peak for many regions (Figure 5A). For T1w/T2w, more regions were shown to peak before the maximum age of 88.9, but the onset decline age was largely comparable to quadratic peak age (Figure 5B).

Discussion

Both MTR and T1w/T2w, although limited by their qualitative nature, were able to capture meaningful intracortical myelin trajectory features across the lifespan. Consistent with previous findings, we observed an inverted-U shape trajectory of intracortical myelination for most of the cerebral cortex. While the regional peak ages were correlated between measures, T1w/T2w increases were found to be more protracted than MTR. This may reflect potential contributions to T1w/T2w from iron content, which naturally occurs in the brain and accumulates throughout life [18].

Frontal and temporal lobe regions showed consistently later peak ages compared with parietal and sensory motor regions. Differences in relative peak age between measures were observed in occipital and insular areas and may potentially reflect bias fields that persisted after taking the ratio of images to calculate MTR and T1w/T2w.

Using a spline fitting, more features of the myelination trajectory were identified for both measures. The wide age range, especially on the aging side, allowed us to distinguish an onset decline age in a larger number of regions than previously reported [5]. Overall, T1w/T2w appeared to be more accurate in relative myelin distribution across regions while MTR showed trajectories that are more in line with existing literature.

Acknowledgements

This study was supported in part by a Developmental Project Grant awarded by the NYU Langone Alzheimer’s Disease Research Center supported by the National Institute on Aging (NIA) grants P30AG066512. Data used here was provided by the Cambridge Centre for Ageing and Neuroscience (Cam-CAN). Cam-CAN funding was supported by the UK Biotechnology and Biological Sciences Research Council (grant number BB/H008217/1), together with support from the UK Medical Research Council and University of Cambridge, UK.