Characterizing longitudinal white matter development during early childhood

Abstract

Post-mortem studies have shown the maturation of the brain's myelinated white matter, crucial for efficient and coordinated brain communication, follows a nonlinear spatio-temporal pattern that corresponds with the onset and refinement of cognitive functions and behaviors. Unfortunately, investigation of myelination in vivo is challenging and, thus, little is known about the normative pattern of myelination, or its association with functional development. Using a novel quantitative magnetic resonance imaging technique sensitive to myelin we examined longitudinal white matter development in 108 typically developing children ranging in age from 2.5 months to 5.5 years. Using nonlinear mixed effects modeling, we provide the first in vivo longitudinal description of myelin water fraction development. Moreover, we show distinct male and female developmental patterns, and demonstrate significant relationships between myelin content and measures of cognitive function. These findings advance a new understanding of healthy brain development and provide a foundation from which to assess atypical development.

Fig. 1

Top row: ages of the 108 subjects at each scan. Each row denotes an individual subject and the repeated measurements are connected with a dashed line. Males (blue) and females (green) are additionally distinguished. Bottom row: mean mcDESPOT VF_M maps at 3, 6, 9, 12, 24, 36, 48, and 60 months of age illustrating the development of myelinated white matter throughout the brain

Fig. 2

Representative mean VF_M developmental trajectories. Repeated measurements for each subject are connected with a straight line. Developmental trajectories are observed to follow a “S” shape pattern that is characteristic to a sigmoidal function. Anatomical locations associated with each graph: A frontal lobe white matter, B caudate, C insula, D putamen, E thalamus, F temporal lobe white matter, G parietal lobe white matter, H occipital lobe white matter, I cerebellar white matter

Fig. 3

Reconstructed modified Gompertz trajectories for each subject (blue curves) and the overall population (black dashed curve) for a representative subset of the investigated regions. Parameters for these models were estimated by fitting the VF_M profile to the modified Gompertz function as a function of gestationally corrected age using nonlinear mixed effects modeling

Fig. 4

Top Row: 95 % confidence intervals for the frontal white matter and body of the corpus callosum. Using the standard error estimates from the nonlinear mixed effects modeling, confidence intervals of the population trajectory can be estimated. Bottom row: representative VF_M growth rate curves. These curves were reconstructed by taking the time derivative of the modified Gompertz function (Eq. 2) and using the overall population estimates of the modified Gompertz parameters. Such curves are informative of the rate of change of the VF_M with respect to age (time) and highlight the posterior–anterior developmental gradient of VF_M

Fig. 5

VF_M developmental trajectory of the splenium of the corpus callosum. Trajectories are separated by gender, highlighting the developmental profile difference between males and females

Fig. 6

Top row: change in visual reception, gross motor, and receptive language Mullen scales of early learning assessment scores as a function of changing VF_M for the thalamus. The change in cognitive assessment scores were found to significantly correlate with changing VF_M, suggesting concomitant structure–function development. Bottom row: illustrative moving average correlations (Pearson’s r, y-axis) as a function of mean age (days, x-axis) for the thalamus. Plots demonstrate the dynamic relationships between ${\mathit{\Delta }}_{{\text{VF}}_{\text{M}}}$ and fine motor and expressive language Δ_M. The age range at which these relationships transition appears to overlap with changes in the VF_M and ∂_t VF_M trajectories