Network-level structural covariance in the developing brain

Brandon A. Zielinski, Efstathios D. Gennatas, Juan Zhou, William W. SeeleyProceedings of the National Academy of Sciences 2010 · DOI

TL;DR

This study uses structural covariance MRI to map how large-scale brain networks form in children. They found that basic sensory and motor networks are established early, while more complex cognitive and language networks mature gradually throughout adolescence, revealing distinct developmental trajectories for different brain systems.

What Problem Does This Paper Address?

Understanding how the complex network architecture of the adult brain emerges during development is a significant challenge. While functional connectivity MRI (fMRI) has identified distinct networks in adults, and some preliminary work exists for infants, the developmental trajectory of these large-scale networks throughout childhood and adolescence remains largely unknown. This gap hinders our understanding of how brain function is assembled and how developmental disruptions might lead to neurological disorders. This paper aims to fill this gap by characterizing the development of structural covariance networks (SCNs) across four age groups of children, hypothesizing that networks supporting more complex cognitive functions will show a more protracted maturation.

Methods

The researchers employed structural covariance MRI (scMRI) to investigate relationships between gray matter volume across different brain regions. This technique assumes that if two brain regions consistently show similar gray matter volume across individuals, they are likely part of the same functional network.

They used a seed-based approach:

  1. Seed Selection: They identified eight "seed" regions in the brain, each known to be a key node in well-established functional networks (e.g., visual cortex, auditory cortex, motor cortex, language areas, salience network, executive control network, and the default-mode network). These seeds were based on prior fMRI studies.
  2. Whole-Brain Covariance Mapping: For each seed region and each age group, they calculated the correlation in gray matter volume between the seed and every other voxel in the brain. This generated a "structural covariance network" (SCN) for each seed and age group.
  3. Statistical Thresholding: The resulting covariance maps were thresholded using a stringent statistical criterion (P < 0.001, family-wise error corrected) to identify significant structural relationships.
  4. Quantitative Analysis: To track developmental changes, they quantified the spatial extent of each SCN by counting the number of significant voxels. These counts were then plotted across the four age groups to visualize developmental trajectories.
  5. Trajectory Modeling: Follow-up analyses used statistical models to formally classify SCN trajectories as either "peak and prune" (developed early, then reduced) or "cumulative growth" (gradually increasing over time).

Datasets

The study utilized data from the National Institutes of Health (NIH) MRI Study of Normal Brain Development's Pediatric MRI Data Repository. This repository contains T1-weighted MRI scans of 432 normal children aged 4.5 to 21 years. The researchers selected a subset of 300 children, divided into four sex-matched age groups:

  • Group 1: Early childhood (4.9-7.8 years)
  • Group 2: Late childhood (8.5-11.2 years)
  • Group 3: Early adolescence (12.1-14.2 years)
  • Group 4: Late adolescence (16.0-18.0 years)

This dataset is a valuable resource for studying brain development and differs from typical adult datasets by focusing specifically on a wide age range of children.

Key Insights and Ideas

  • Domain-Specific Developmental Trajectories: The most striking finding is that different large-scale brain networks follow distinct developmental paths. This suggests that brain maturation is not a uniform process but rather a system-specific assembly.
  • Early Maturation of Sensory/Motor Systems: Networks anchored by primary sensory and motor cortices are surprisingly well-developed even in young children. This aligns with the idea that basic perceptual and motor functions are foundational and mature earlier.
  • Gradual Maturation of Higher-Order Networks: Language, social-emotional, and executive control networks show a much slower, progressive expansion in their structural covariance throughout childhood and adolescence. This highlights the protracted development of complex cognitive functions.
  • "Peak and Prune" vs. "Cumulative Growth": The study provides evidence for two main developmental patterns:
  • Peak and Prune: Sensory and motor networks showed an expansion in early adolescence followed by a reduction in late adolescence, suggesting a refinement or pruning process.
  • Cumulative Growth: Higher-order networks showed continuous expansion, indicating ongoing development and integration.
  • Default Mode Network (DMN) Nuance: The DMN, often associated with self-referential thought and mind-wandering, showed a trajectory similar to sensory/motor networks (expansion then potential pruning), but with some limitations in its full adult-like connectivity even in late adolescence. This suggests its complex structure might involve subnetworks with different developmental timelines.
  • SCNs as a Window into Development: The study reinforces the utility of SCNs as a measure of network integrity and development, mirroring findings from functional connectivity studies.

Key Results

  • Primary Sensory and Motor Networks: These networks were present in early childhood and showed significant expansion in early adolescence (Group 3) before becoming more restricted in late adolescence (Group 4), fitting a "peak and prune" trajectory.
  • Language-Related Networks (Speech and Semantic): These networks showed limited development in younger children and progressively increased in regional extent and distribution throughout adolescence, fitting a "cumulative growth" trajectory.
  • Salience and Executive Control Networks: Similar to language networks, these showed limited early covariance and significant expansion throughout childhood and adolescence, also fitting a "cumulative growth" trajectory.
  • Default Mode Network (DMN): This network showed a peak in spatial extent in Group 3 (early adolescence) and then a restriction in Group 4, similar to sensory/motor networks, but with some missing connections to frontal regions seen in adult DMNs.
  • Sex Differences: Preliminary analyses suggested subtle but compelling differences, with females showing earlier expansion in language and salience networks, consistent with existing literature suggesting earlier maturation in these domains for females.

Limitations and Future Work

  • Cross-Sectional Design: The study is cross-sectional, meaning it compares different groups of children at different ages rather than tracking the same individuals over time. Longitudinal studies are needed to confirm these developmental trajectories.
  • DMN Complexity: The DMN's trajectory was somewhat unexpected and may reflect the influence of multiple subnetworks, some of which might not have been captured by the chosen seed regions or analysis methods. Future work could explore DMN subnetworks in more detail.
  • Laterality and Sex Effects: While preliminary analyses were conducted, the study was not specifically designed to address laterality or sex differences in detail. The authors suggest that future studies with larger sample sizes and specific designs are needed to fully explore these nuances.
  • Relationship between ICN and SCN: The biological underpinnings of SCNs (e.g., trophic influences, axonal guidance) and their precise relationship with functional connectivity (ICNs) remain an area for future investigation. Multimodal approaches combining structural and functional data are recommended.
  • Broader Implications: The authors suggest that understanding these developmental trajectories is crucial for interpreting neurodevelopmental disorders and age-related brain changes. Future research could use these findings to inform diagnostic and therapeutic strategies.

Summary updated

Abstract

Intrinsic or resting state functional connectivity MRI and structural covariance MRI have begun to reveal the adult human brain's multiple network architectures. How and when these networks emerge during development remains unclear, but understanding ontogeny could shed light on network function and dysfunction. In this study, we applied structural covariance MRI techniques to 300 children in four age categories (early childhood, 5-8 y; late childhood, 8.5-11 y; early adolescence, 12-14 y; late adolescence, 16-18 y) to characterize gray matter structural relationships between cortical nodes that make up large-scale functional networks. Network nodes identified from eight widely replicated functional intrinsic connectivity networks served as seed regions to map whole-brain structural covariance patterns in each age group. In general, structural covariance in the youngest age group was limited to seed and contralateral homologous regions. Networks derived using primary sensory and motor cortex seeds were already well-developed in early childhood but expanded in early adolescence before pruning to a more restricted topology resembling adult intrinsic connectivity network patterns. In contrast, language, social-emotional, and other cognitive networks were relatively undeveloped in younger age groups and showed increasingly distributed topology in older children. The so-called default-mode network provided a notable exception, following a developmental trajectory more similar to the primary sensorimotor systems. Relationships between functional maturation and structural covariance networks topology warrant future exploration.
BibTeX
@inproceedings{zielinski2010network,
  author = {Brandon A. Zielinski and Efstathios D. Gennatas and Juan Zhou and William W. Seeley},
  title = {{Network-level structural covariance in the developing brain}},
  booktitle = {Proceedings of the National Academy of Sciences},
  year = {2010},
  doi = {10.1073/pnas.1003109107},
}
Figure