Ageing and brain white matter structure in 3,513 UK Biobank participants

Abstract

Quantifying the microstructural properties of the human brain's connections is necessary for understanding normal ageing and disease. Here we examine brain white matter magnetic resonance imaging (MRI) data in 3,513 generally healthy people aged 44.64-77.12 years from the UK Biobank. Using conventional water diffusion measures and newer, rarely studied indices from neurite orientation dispersion and density imaging, we document large age associations with white matter microstructure. Mean diffusivity is the most age-sensitive measure, with negative age associations strongest in the thalamic radiation and association fibres. White matter microstructure across brain tracts becomes increasingly correlated in older age. This may reflect an age-related aggregation of systemic detrimental effects. We report several other novel results, including age associations with hemisphere and sex, and comparative volumetric MRI analyses. Results from this unusually large, single-scanner sample provide one of the most extensive characterizations of age associations with major white matter tracts in the human brain.

Figure 1. White matter tracts of interest.

Generated using probabilistic tractography rendered in superior (left), anterior (centre) and lateral (right) views.

Figure 2. Age associations with the microstructural characteristics of association fibres.

(a) Kernel density plots indicate the degree of data point overlap (darker=greater); black line denotes the linear or quadratic regression line (with grey 95% CIs) across the five microstructural measures. Blue, FA; green, MD; red, intracellular volume fraction; purple, ISOVF; orange, OD. G, gyrus; IFOF: inferior fronto-occipital fasciculus; ILF, inferior longitudinal fasciculus; PH, parahippocampal; SLF, superior longitudinal fasciculus.

Figure 3. Age associations with the microstructural characteristics of thalamic and callosal fibres.

Kernel density plots indicate the degree of data point overlap (darker=greater); black line denotes the linear or quadratic regression line (with grey 95% CIs) across the five microstructural measures. Blue, FA; green, MD; red, intracellular volume fraction; purple, ISOVF; orange, OD. ATR, anterior thalamic radiation; PTR, posterior thalamic radiation; STR, superior thalamic radiation.

Figure 4. Age associations with the microstructural characteristics of sensory projection fibres.

Kernel density plots indicate the degree of data point overlap (darker=greater); black line denotes the linear or quadratic regression line (with grey 95% CIs) across the five microstructural measures. Blue, FA; green, MD; red, intracellular volume fraction; purple, ISOVF; orange: OD.

Figure 5. Associations between diffusion parameters with age, sex and hemisphere.

FA (blue), MD (green), ICVF (red), ISOVF (purple) and OD (orange) with age, sex and hemisphere. Error bars=95% CIs. Female and left hemisphere coded as 0. The valence of MD and ISOVF associations have been reflected for the purposes of visualization for all four panels. See Supplementary Tables 3 and 4 for regression coefficients.

Figure 6. Association magnitudes of age and tract averaged MD.

Rendered in superior (left), anterior (centre) and lateral (right) views. Coefficient values (standardized βs) are from linear components of models shown in Supplementary Tables 3 and 4.

Figure 7. Illustrative heatmaps of tract de-differentiation for each parameter across age groups.

All tracts are shown. Higher tract inter-correlations are indicated by oranges and darker reds, with blues and greens denoting lower magnitudes.

Figure 8. General factors of white matter microstructure explain greater variance with increasing age.

(a) Scree slopes for the exploratory factor analysis, showing the eigenvalue against the number of factors for each white matter tract measurement. (b) Age trajectories of the first (latent) factor of white matter microstructure for each of the five dMRI biomarkers. (c) Age de-differentiation of white matter microstructure. Age trajectories for the proportion of total variance in each tract measurement explained by the general factor. The shaded region around each trajectory shows ±1 s.d. of the mean.