The first three years of life are a period of rapid brain development, marked by increasing refinement of cortical organization and the emergence of large-scale brain systems resembling their adult configurations. This process involves increased cortical network segregation, where brain regions show denser interconnections within systems and greater distinction between systems. Maturation occurs in a spatiotemporal pattern, starting earlier in primary sensory areas than in association areas. The level of cortical network segregation is linked to cognitive abilities in adolescents and adults, with higher levels often associated with better performance. One environmental factor significantly impacting this development, along with many other life outcomes, is socioeconomic disadvantage, a multifaceted construct encompassing socioeconomic status (SES) indicators such as income, education, and occupation, as well as health factors like insurance and nutrition. Previous research, mainly focusing on older children and adolescents, suggests that disadvantage may accelerate cortical network segregation. However, no longitudinal studies have examined this relationship during the crucial first three years of life. Understanding the impact of early disadvantage on the pace of brain development is vital because changes in this pace have been linked to psychiatric disorders and accelerated pubertal development. An accelerated pace might lead to earlier declines in brain plasticity, hindering the development of optimal cortical circuitry adapted to the environment. This research tests the hypothesis that prenatal disadvantage is associated with differences in the pace of functional brain network development, specifically predicting that children from more advantaged backgrounds will show a more protracted trajectory of cortical network segregation, potentially linked to extended experience-dependent learning and brain plasticity.

Existing literature demonstrates a strong link between early environments and brain structure and function. Studies in youth aged 8-22 years show that socioeconomic status (SES) moderates age-associated increases in cortical network segregation, with more advantaged youth exhibiting a steeper increase but starting with lower initial segregation. Similar associations are found in youth aged 6-17 years living in low-SES neighborhoods. These findings suggest that disadvantage might accelerate the pace of cortical network segregation early in development. While early-life associations between disadvantage and brain organization are evident as early as the first month of life, longitudinal studies investigating the relationship with cortical network segregation development during the first few years of life are lacking. Theoretical models, drawing from evolutionary developmental biology, propose that early environments influence maturation pace, with harsh environments possibly leading to accelerated development and nurturing environments resulting in a more prolonged developmental trajectory. However, the precise nature of this relationship and its impact on intrinsic brain network development requires further investigation.

This study leverages a longitudinal neuroimaging dataset from the Early Life Adversity, Biological Embedding, and Risk for Developmental Precursors of Mental Health Disorders (eLABE) cohort, comprising neonates and toddlers with extensively characterized early environments. Prenatal disadvantage was assessed using a latent factor derived from maternal income-to-needs ratio, educational attainment, area deprivation index, insurance status, and nutrition. The study included 261 participants at the neonatal timepoint, with smaller sample sizes at the 2-year (n=92) and 3-year (n=66) timepoints due to COVID-related data collection limitations. Resting-state fMRI data were acquired at each timepoint using a 3T Prisma scanner. Preprocessing involved correcting for intensity differences, bias field correction, motion correction, and registration to the Talairach atlas. For neonates, BOLD images were registered to individual T2-weighted images, and then to a group-average T2 from the cohort, finally registering to the Talairach atlas. For toddlers, BOLD images were registered to individual T1 weighted images, then to a group-average T1, and finally to the Talairach atlas. Head motion was carefully addressed using FIRMM monitoring and censoring, applying different thresholds for neonates and toddlers. A minimum of 5 minutes of data post-censoring was needed for inclusion. Network analysis was performed using a 333-region cortical parcellation, calculating measures of global, meso-scale, and local segregation (system segregation, modularity, and clustering coefficient, respectively), as well as network integration (participation coefficient). Generalized Additive Mixed Models (GAMMs) were used to model the developmental trajectories of these network measures, incorporating age as a smooth term, sex, motion metrics, number of retained fMRI frames, and average network connectivity as covariates. Interactions between prenatal disadvantage and age were included to test the effect of disadvantage on the pace of development. Regional specificity was explored using region-specific GAMMs. Linear regression models, accounting for relevant covariates, examined associations between network segregation at ages 2 and 3 and language and cognitive composite scores from the Bayley-III. Sensitivity analyses explored the robustness of findings to various methodological variations and potential confounds.

Cortical network segregation increased significantly with age across all scales (global, meso-scale, and local) from birth to 3 years. The increase was most pronounced during the first two years. Prenatal disadvantage significantly moderated the trajectories of cortical network segregation at all scales, with infants and toddlers experiencing greater prenatal disadvantage exhibiting a faster increase in segregation. These age-by-disadvantage interactions were strongest for local segregation (clustering coefficient). When controlling for local segregation, age-by-disadvantage effects on global and meso-scale segregation were no longer significant. Regionally, age-by-disadvantage associations with local segregation were widespread (56% of regions), with strongest associations in somatomotor and dorsal attention systems, aligning with a sensorimotor-association cortical hierarchy. The magnitude of age-by-disadvantage effects on local segregation was negatively correlated with sensorimotor-association axis ranks. At age 2, whole-cortex local segregation was negatively associated with language scores (t(83)=-2.45, p=0.017, PFDR = 0.033), this association remained significant (t(83)=-2.33, p = 0.022) even after controlling for prenatal disadvantage. Similar negative correlations were observed between local segregation and language scores at age 3. There was no significant association between segregation and cognitive scores. Sensitivity analyses confirmed the robustness of the main findings to various methodological changes and controls for potential confounds, including using a different measure of SES (income-to-needs ratio and maternal education) and addressing sample differences over the study period, head motion, changes in disadvantage over time, longitudinal modeling choices, or outlier effects. Finally, no significant associations were found between prenatal psychosocial stress and developmental changes in cortical network segregation.

This study demonstrates a strong association between prenatal disadvantage and the pace of cortical network development during the first three years of life. Children from more disadvantaged backgrounds exhibited accelerated development of cortical network segregation, particularly in early-developing sensorimotor regions. This accelerated development may be linked to earlier declines in cortical plasticity. The findings support a model where early environments influence the tempo of brain development and that a more advantaged environment may lead to more protracted cortical functional network development. The negative association between local segregation and language abilities at age 2, even after controlling for prenatal disadvantage, suggests a potentially independent link between network architecture and language development. This contrasts with the typically observed positive association between segregation and cognitive performance in older individuals, underscoring the importance of developmental context in interpreting such relationships. The observed associations with somatomotor regions might reflect either the long-lasting effects of early environment or the greater sensitivity of sensorimotor systems to environmental effects in combination with lower interindividual variability. Future research should further explore this and investigate the precise mechanisms linking environmental factors and maturational pace.

This study provides compelling evidence that prenatal disadvantage accelerates cortical network segregation during early development. This accelerated development is particularly prominent in sensorimotor regions and is associated with language abilities. The findings highlight the importance of early interventions to mitigate the impact of disadvantage on brain development and underscore the need for policies supporting parents of young children. Future research should examine the interplay of different environmental factors with brain development, utilize infant-specific parcellations, and extend the longitudinal follow-up to middle childhood to fully understand the long-term consequences of early developmental trajectories.

The study's smaller sample sizes at the 2- and 3-year timepoints due to the COVID-19 pandemic limit the statistical power of the behavioral analyses, which should be considered exploratory. The use of an adult parcellation for neonatal data is a potential limitation, though it was chosen as it is the only way to align data across different ages. Future research should incorporate infant-specific parcellations. The study focuses on socioeconomic disadvantage, but other forms of adversity, such as discrimination and racism, should be investigated in future studies. The measurement of psychosocial stress was limited, so the study could not fully untangle its individual contribution to the observed effects. Finally, while the study finds an association between accelerated brain development and language abilities, further investigation is needed to determine if similar relationships exist for cognition and psychopathology.