Social-Cognitive Network Connectivity in Preterm Children and Relations With Early Nutrition and Developmental Outcomes

Abstract

Infants born very low birth weight (VLBW, < 1,500 g) are at a heightened risk for structural brain abnormalities and social-cognitive deficits, which can impair behavioural functioning. Resting-state fMRI, reflecting a baseline level of brain activity and underlying social-cognitive processes, has also been reported to be altered in children born VLBW. Yet very little is known about the functional networks underlying social cognition using magnetoencephalography (MEG) and how it relates to neonatal factors and developmental outcomes. Thus, we investigated functional connectivity at rest in VLBW children and the associations with early nutrition and IQ and behavioural problems. We collected resting-state MEG recordings and measures of IQ and social-cognitive behaviour, as well as macronutrient/energy intakes during initial hospitalisation in 5-year-old children born VLBW (n = 37) compared to full-term (FT; n = 27) controls. We examined resting-state network differences controlling for sex and age at scan. Functional connectivity was estimated using the weighted phase lag index. Associations between functional connectivity with outcome measures and postnatal nutrition were also assessed using regression analyses. We found increased resting-state functional connectivity in VLBW compared to FT children in the gamma frequency band (65-80 Hz). This hyper-connected network was primarily anchored in frontal regions known to underlie social-cognitive functions such as emotional processing. In VLBW children, increased functional connectivity was related to higher IQ scores, while reduced connectivity was related to increased behavioural problems at 5 years of age. These within-group associations were found in the slower frequency bands of theta (4-7 Hz) and alpha (8-12 Hz), frequently linked to higher-order cognitive functions. We also found significant associations between macronutrient (protein and lipid) and energy intakes during the first postnatal month with functional connectivity at preschool-age, highlighting the long-term impacts of postnatal nutrition on preterm brain development. Our findings demonstrate that at preschool-age, VLBW children show altered resting-state connectivity despite IQ and behaviour being in the average range, possibly reflecting functional reorganisation of networks to support social-cognitive and behavioural functioning. Further, our results highlight an important role of early postnatal nutrition in the development of resting-state networks, which in turn may improve neurodevelopmental outcomes in this vulnerable population.

Keywords: MEG (magnetoencephalography); functional connectivity; nutrition; outcomes; preterm (birth); resting-state; social-cognition; very low birth weight.

FIGURE 1

Between-group network analysis. Increased resting-state functional connectivity in VLBW compared to FT children in the gamma (65–80 Hz) frequency band (42 edges, 29 nodes, p_corr = 0.03). Node size is scaled by degree (i.e., the number of connections a node has).

FIGURE 2

Within-group network analysis. Significant alpha-band (8–14 Hz) network positively associated with IQ in VLBW children (40 edges, 33 nodes, p_corr = 0.033). Node size is scaled by degree.

FIGURE 3

Within-group network analysis. (A) Theta-band (4–7 Hz) network significantly associated with BASC-3 externalising problems (44 edges, 36 nodes, p_corr = 0.039) in VLBW children. (B) Alpha-band (8–14 Hz) network significantly associated with BASC-3 internalising problems (41 edges, 32 nodes, p_corr < 0.001). (C) Alpha-band network significantly associated with the BASC-3 behavioural symptoms index (38 edges, 33 nodes, p_corr = 0.002). All of these were negative correlations, such that reduced network connectivity was related to increased behavioural difficulties. Node size is scaled by degree.

FIGURE 4

Within-group network analysis: VLBW group. (A) Significant alpha-band (8–14 Hz) network positively associated with higher protein intake (40 edges, 31 nodes, p_corr = 0.032) in VLBW children. (B) Significant alpha-band (8–14 Hz) network positively associated with energy intake (40 edges, 36 nodes; p_corr = 0.034). (C) Significant beta-band (15–29 Hz) negatively associated with lipid intake (41 edges, 37 nodes, p_corr = 0.007). Node size is scaled by degree.