Dissociation of reliability, heritability, and predictivity in coarse- and fine-scale functional connectomes during development

Erica L. Busch, Kristina M. Rapuano, Kevin M. Anderson, Monica D. Rosenberg, Richard Watts, B.j. Casey, James V. Haxby, Ma Feilong

TL;DR

This study reveals that different spatial scales of brain connectivity (coarse vs. fine) are associated with distinct aspects of individual differences in children. While coarse-scale connectivity is more heritable, fine-scale connectivity is more reliable and better predicts cognitive abilities, suggesting a dissociation between genetic influences and individual experiences in shaping brain function.

What Problem Does This Paper Address?

Understanding how brain structure and function relate to individual differences in cognition is a central goal in neuroscience. Traditionally, studies have focused on "coarse-grained" measures of brain connectivity, often by averaging signals across large brain regions. While this approach has yielded insights, it might obscure finer-scale details that could be crucial for understanding individual variability, especially during development. The authors address the question of whether these different spatial scales of brain connectivity (coarse vs. fine) are differentially influenced by genetics and experience, and how they relate to cognitive abilities in children. They hypothesize that coarse-scale connectivity might be more heritable, reflecting fundamental genetic blueprints, while fine-scale connectivity might be more influenced by individual experiences and thus more reliable and predictive of behavior.

Methods

The researchers employed Connectivity Hyperalignment (CHA), a machine learning technique that aligns functional connectivity patterns across individuals in a shared "common space." This method is crucial for overcoming anatomical variability and improving the reliability of fine-scale (vertex-wise) connectivity measures. They compared CHA-aligned connectomes with traditionally "anatomically aligned" (AA) connectomes, which rely on aligning brains based on their physical structure.

They analyzed two levels of connectome granularity:

  • Fine-scale connectomes: Represent connectivity between individual vertices (small points on the brain surface) within a parcel and the average signal of other parcels.
  • Coarse-scale connectomes: Represent connectivity between parcels (larger brain regions), where signals within each parcel are averaged.

Key analyses included:

  • Reliability analysis: Assessed how consistently individual differences in connectivity profiles could be measured within subjects using split-half resting-state fMRI data.
  • Heritability analysis: Used data from twin pairs to estimate the genetic contribution to connectivity at different scales and alignments.
  • Neurocognition PC prediction analysis: Used ridge regression to predict individual differences in cognitive abilities from connectome data.

Datasets

The study utilized data from the Adolescent Brain Cognitive Development (ABCD) Study, a large, longitudinal study of children's brain development. Specifically, they used resting-state fMRI and behavioral data from a subset of 1,115 children aged 9-10 years. This cohort included monozygotic (MZ) and dizygotic (DZ) twin pairs, as well as unrelated individuals, providing a rich dataset for examining genetic and environmental influences.

Key Insights and Ideas

  • Dissociation of Scales: The study strongly suggests that coarse- and fine-scale brain connectivity capture fundamentally different types of information.
  • Reliability vs. Heritability Trade-off: Fine-scale connectivity, while more reliable, is less heritable, implying a greater role for individual experiences. Conversely, coarse-scale connectivity, though less reliable, is more strongly influenced by genetics.
  • Hyperalignment's Power: CHA significantly improves the reliability of fine-scale connectivity, making it a powerful tool for developmental neuroscience.
  • Cognitive Relevance: Different connectome scales predict different cognitive traits. Highly heritable traits are better predicted by heritable (coarse-scale) connectomes, while less heritable traits are better predicted by reliable, experience-driven (fine-scale) connectomes.

Key Results

  • Reliability: CHA significantly increased the reliability of individual differences in functional connectivity at both coarse and fine scales. Fine-scale connectomes were consistently more reliable than coarse-scale connectomes after CHA (average r = 0.86 for fine vs. 0.58 for coarse).
  • Heritability: Coarse-scale connectomes were significantly more heritable (mean h²-multi = 0.18) than fine-scale connectomes (mean h²-multi = 0.10). Specifically, anatomically aligned coarse-scale connectomes showed the highest heritability (0.20 ± 0.05).
  • Prediction:
    • General cognitive ability (a highly heritable trait) was similarly predicted by CHA fine, AA coarse, and AA fine connectomes. However, CHA coarse connectomes predicted significantly less variance.
  • Learning/memory (a less heritable trait) was best predicted by the CHA fine connectome, outperforming other connectome types.

Limitations and Future Work

  • Low Prediction Variance: The authors acknowledge that their prediction models accounted for relatively low variance in cognitive measures compared to adult studies. This is attributed to the inherent stability of resting-state fMRI and the challenges of collecting high-quality data in children.
  • Task-Based vs. Resting-State FC: While resting-state FC is more heritable, task-based FC is generally more stable and behaviorally predictive. Future work could explore task-based FC in children.
  • Longitudinal Analysis: The study used cross-sectional data. Future work could leverage the longitudinal nature of the ABCD study to investigate how heritability and predictivity of connectivity change over time.
  • Age and Diversity: While the study included a diverse cohort, future research could further investigate how heritability scores and brain-based prediction vary with age and across different demographic groups.
  • Functional Topographies and Development: The authors suggest that their approach could be extended to investigate how functional topographies reflect cognitive development over time by hyperaligning subjects to themselves at multiple timepoints.

Summary updated

Abstract

Abstract The functional connectome supports information transmission through the brain at various spatial scales, from exchange between broad cortical regions to finer–scale, vertex–wise connections that underlie specific information processing mechanisms. In adults, while both the coarse- and fine-scale functional connectomes predict cognition, the fine-scale can predict up to twice the variance as the coarse-scale functional connectome. Yet, past brain-wide association studies, particularly using large developmental samples, focus on the coarse connectome to understand the neural underpinnings of individual differences in cognition. Using a large cohort of children (age 9 – 10 years; n = 1,115 individuals, both sexes, 50% female, including 170 monozygotic and 219 dizygotic twin pairs and 337 unrelated individuals), we examine the reliability, heritability, and behavioral relevance of resting-state functional connectivity computed at different spatial scales. We use connectivity hyperalignment to improve access to reliable fine-scale (vertex–wise) connectivity information and compare the fine-scale connectome with the traditional parcel–wise (coarse scale) functional connectomes. Though individual differences in the fine-scale connectome are more reliable than those in the coarse-scale, they are less heritable. Further, the alignment and scale of connectomes influence their ability to predict behavior, whereby some cognitive traits are equally well predicted by both connectome scales, but other, less heritable cognitive traits are better predicted by the fine-scale connectome. Together, our findings suggest there are dissociable individual differences in information processing represented at different scales of the functional connectome which, in turn, have distinct implications for heritability and cognition. Significance statement Years of human magnetic resonance imaging (MRI) research demonstrate that individual variability in resting-state functional connectivity relates to genetics and cognition. However, the various spatial scales where individual differences in connectivity could occur have yet to be considered in childhood brain– behavior association studies. Here, we use novel machine learning approaches to examine the reliability, heritability, and behavioral relevance of different spatial scales of the resting-state functional connectome during childhood. We show that broad features of the connectome are strongly related to heritability, whereas fine details are more reliable and strongly associated with neurocognitive performance. These data indicate that reliable, heritable, and behaviorally–relevant individual differences exist at dissociable scales of the functional connectome.
BibTeX
@article{busch2023dissociation,
  author = {Erica L. Busch and Kristina M. Rapuano and Kevin M. Anderson and Monica D. Rosenberg and Richard Watts and B.J. Casey and James V. Haxby and Ma Feilong},
  title = {{Dissociation of reliability, heritability, and predictivity in coarse- and fine-scale functional connectomes during development}},
  journal = {Journal of Neuroscience},
  year = {2023},
  doi = {10.1523/jneurosci.0735-23.2023},
}
Figure