Gray and white matter functional organization and their individual variabilities in normal aging

1495 

Abstract Submission 

Chenye Shen¹, Chenlu Ma², Chaoqiang Liu¹, Nanguang Chen¹, Anqi Qiu^1,3,4

¹Department of Biomedical Engineering, National University of Singapore, Singapore, Singapore, ²Department of Chinese Language and Literature, Fudan University, Shanghai, China, ³Department of Health Technology and Informatics, Hong Kong Polytechnic University, Kowloon, Hong Kong, ⁴Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD

Chenye Shen  
Department of Biomedical Engineering, National University of Singapore
Singapore, Singapore

Chenlu Ma  
Department of Chinese Language and Literature, Fudan University
Shanghai, China

Chaoqiang Liu  
Department of Biomedical Engineering, National University of Singapore
Singapore, Singapore

Nanguang Chen  
Department of Biomedical Engineering, National University of Singapore
Singapore, Singapore

Anqi Qiu  
Department of Biomedical Engineering, National University of Singapore|Department of Health Technology and Informatics, Hong Kong Polytechnic University|Department of Biomedical Engineering, Johns Hopkins University
Singapore, Singapore|Kowloon, Hong Kong|Baltimore, MD

The normal aging brain experiences a progressive deterioration in brain function, impacting cognition (Hedden et al. 2004). Recent studies reveal that brain function is embedded in a large-scale functional organization. While prior studies focused on gray matter (GM) functional organization in the cortex (Margulies et al. 2016), cerebellum (Guell et al. 2018), and subcortex (Tian et al. 2020), emerging evidence underscores the importance of BOLD signals in white matter (WM) (Peer et al. 2017). A recent study uncovered WM and GM function organization interaction in neurodevelopment (Zhu et al. 2023). This study explores the interplay between GM and WM functional organization in a normal aging population, hypothesizing that WM functional organization alterations align with GM changes and are linked to cognitive performance.

This study leveraged demographic, cognitive function tests, disease diagnosis, and brain imaging data (T1 and rs-fMRI) from the UK Biobank. We excluded participants with cancers, neurological or cardiovascular diseases, resulting in a normal aging population (N=23,051, age=44~82 years). T1 and rs-fMRI data were preprocessed in FreeSurfer and FSL. Cortical GM and WM functional organizations were identified using methods from (Zhu et al. 2023). Group-level GM and WM masks were first generated based on T1 segmentation (Peer et al. 2017), and GM and WM functional gradients were computed separately through the decomposition of corresponding functional connectivity (FC) (Margulies et al. 2016). Cortical GM was parcellated into 7 networks (Yeo et al. 2011). The WM functional networks were identified by assessing the voxel's highest FC strength with GM, associating the WM network with the functional role of its corresponding GM counterpart.
We assessed the individual variability of each gradient by calculating the Euclidian distance between individual and group-averaged gradient space. The age-related individual variability changes were investigated using linear regression. We also assessed its cognitive relevance, covarying for age, sex, and education.

Figure 1 illustrates GM and WM functional gradients. The first GM gradient explained 20.6% of cortical GM FC variance, with limbic (Lim) and default mode network (DMN) anchoring the transmodal pole and visual (Vis) and sensorimotor (SM) networks at the unimodal end (Fig. 1a). The second GM gradient, accounting for 16.8% of variance, was situated between Vis and SM networks (Fig. 1b). The first two WM gradients, contributing 18.0% and 14.5% to cortical WM FC variance, exhibited a reciprocal mirror ordering, wherein the first WM gradient resembled the second GM gradient, and vice versa (Fig. 1c,d). This suggests that different tissues may preferentially support specific aspects of brain functional organization.
Figure 2 depicts the associations of individual variabilities in gradients with age (Fig. 2a) and six domains of cognitive function (Fig. 2b). It highlights that the ability to capture age and cognitive information was more pronounced in the individual variabilities observed in GM gradients compared to WM, and in the unimodal-transmodal axis compared to the Vis-SM axis. This aligns with fMRI's known sensitivity to GM signals (Gawryluk et al. 2014) and unimodal-transmodal axis' ability to capture diverse brain properties (Bernhardt et al. 2022). The greatest relevance of individual variability to age and cognition at the transmodal end is observed in DMN for GM and Lim for WM, indicating non-uniformities in brain organization (Sydnor et al. 2021).

This study delineated WM functional organization in normal aging, revealing a close alignment with GM functional organization. The functional organization of GM and WM proves effective in capturing age and cognition information in different networks. This work sheds light on future studies investigating WM functional organization and its role in aging.

Executive Function, Cognitive Control and Decision Making

Working Memory

Aging ²

Connectivity (eg. functional, effective, structural) ¹

BOLD fMRI

Aging

Cognition

Data analysis

FUNCTIONAL MRI

White Matter

Other - functional gradients