Infants' cortex undergoes microstructural growth coupled with myelination during development

Abstract

Development of cortical tissue during infancy is critical for the emergence of typical brain functions in cortex. However, how cortical microstructure develops during infancy remains unknown. We measured the longitudinal development of cortex from birth to six months of age using multimodal quantitative imaging of cortical microstructure. Here we show that infants' cortex undergoes profound microstructural tissue growth during the first six months of human life. Comparison of postnatal to prenatal transcriptomic gene expression data demonstrates that myelination and synaptic processes are dominant contributors to this postnatal microstructural tissue growth. Using visual cortex as a model system, we find hierarchical microstructural growth: higher-level visual areas have less mature tissue at birth than earlier visual areas but grow at faster rates. This overturns the prominent view that visual areas that are most mature at birth develop fastest. Together, in vivo, longitudinal, and quantitative measurements, which we validated with ex vivo transcriptomic data, shed light on the rate, sequence, and biological mechanisms of developing cortical systems during early infancy. Importantly, our findings propose a hypothesis that cortical myelination is a key factor in cortical development during early infancy, which has important implications for diagnosis of neurodevelopmental disorders and delays in infants.

Fig. 1. Primary sensory cortices are not fully developed at birth but show extensive microstructural tissue growth during the first 6 months of life.

a Right hemisphere sagittal (top) and ventral-temporal (bottom) T₁ maps in units of seconds [s] displayed on an inflated cortical surface of an example infant across time. Left to right: cortical T₁ at 8 days (newborn), 104 days (~3 months), and 181 days (~6 months) of age (red/yellow: higher T₁; purple: lower T₁). b Primary sensory-motor areas: V1, A1, M1, and S1 shown on the cortical surface of this infant. c T₁ distributions across voxels of each area show a leftward shift from newborns (darker colors) to 6-month-olds (lighter colors). Solid lines: mean distribution; Shaded region: standard error of the mean across 10 infants at each timepoint. NB: newborn; 3 mos: 3-month-old; 6 mos: 6-month-old. Darker colors indicate younger infants. d T₁ linearly decreases with age in primary sensory-motor areas. Each dot: mean T₁ per area per infant. Line: Linear mixed model line fit. e–f Same as in c,d for mean diffusivity (MD). Shaded portions represent 95% confidence intervals. For panels c–f, N_total = 30, 10 infants at each time point (newborn, ~3-month-old, ~6-month-old).

Fig. 2. Hierarchical development of cortical microstructure in visual streams.

a, d Inflated cortical surface of an example 6-month-old infant showing nine dorsal (a) and eight ventral (d) visual areas. b, e T₁ as a function of infant age in each visual area of dorsal (b) and ventral (e) visual processing streams. The color indicates the visual area (see a, d). Each dot: mean T₁ per area per infant. Solid lines: Linear mixed model (LMM) estimates of T₁ development for each visual area. Shaded portions represent 95% confidence intervals. c, f LMM estimates of mean T₁ at birth (LMM intercept) in each dorsal (c) and ventral (f) visual area. Error bars: standard error on estimates of intercepts. Data shown are of the right hemisphere. Left hemisphere data are shown in Supplementary Fig. 5. For panels b, e, N_total = 30, 10 infants at each time point (newborn, ~3-month-old, ~6-month-old).

Fig. 3. Transcriptomic gene analysis of cortical samples reveals that myelination and synaptic processes are cellular mechanisms of postnatal development.

a Matrix showing gene expression levels in prenatal (19 post conceptual weeks (pcw) to 37 pcw) and postnatal (4-months-old) cortical tissue samples for the 95 most differentially expressed genes. N_{prenatal tissue donors} = 7; N_{postnatal tissue donors} = 3. Sample demographics in Supplementary Table 6. Rows: genes, columns: cortical area (acronyms in Supplementary Table 7); color: expression level in reads per kilobase million (RPKM, see colorbar). b Gene expression fold change (FC) between postnatal vs. prenatal cortical samples of the 95 most differentially expressed genes. c Gene enrichment analysis showing the molecular and biological processes and cellular components related to these 95 genes.

Fig. 4. Both R₁ at birth and change in R₁ per day vary systematically across visual processing streams.

Linear mixed modeling (LMM) of R₁ as a function of age (Supplementary Fig. 8) provide estimates of R₁ at birth (LMM intercept) and rate of R₁ development (LMM slope) for each visual area in the dorsal (a, b) and ventral (c, d) visual streams. Error bars: standard error on estimates of intercepts and slopes. Data shown are from right hemisphere; Left hemisphere data are shown in Supplementary Fig. 8.