Distinct neural mechanisms underlying cognitive difficulties in preterm children born at different stages of prematurity

OBJECTIVES: To examine associations between low cognitive-performance and regional-and network-level brain changes at ages 9-10 in very-preterm, moderately-preterm, and full-term children, and explore whether these alterations predict ASD/ADHD symptoms at age 12. METHODS: This longitudinal population-based study included 9-10-year-old U.S. children from ABCD Study. Children underwent brain imaging and cognitive assessment using NIH Toolbox. Cortical thickness and subcortical volumes of preterm-children with low cognitive-performance (NIH composite score < -1SD and > -2SD) were compared with preterm and full-term peers with typical performance (≥-1SD). Structural covariance networks were also examined. RESULTS: Among 7281 children (mean age 9.9 ± 0.6 years; 52.2 % boys), 71 were very-preterm, 151 moderately-preterm, and 7056 full-term. Low cognitive-performance was most prevalent in very-preterm children (29.6 %), followed by moderately-preterm (24.0 %) and full-term children (16.2 %). Very-preterm children with low cognitive-performance had thinner inferior temporal cortex (β = -0.58; p = 0.03), thinner fusiform gyrus (β = -0.62; p = 0.02), and larger amygdala volumes (β = 0.41; p = 0.05) compared to very-preterm children with typical performance. Moderately-preterm children with low cognitive-performance had smaller hippocampal volumes (β = -0.32; p = 0.01). Similar patterns were observed when comparing preterm children with low cognitive-performance to full-term peers with typical performance. Structural covariance network analysis revealed stronger covariance between the precuneus-postcentral gyrus pair among moderately-preterm children with low cognitive-performance. Individualized Differential Structural Covariance Network values extracted from this pair were positively associated with ASD/ADHD symptoms, though not statistically significance. CONCLUSION: Low cognitive performance in preterm children is associated with distinct regional and network-level brain differences, differing by prematurity. Stronger hub covariance may reflect compensatory mechanisms, highlighting the need for prematurity-tailored interventions.<p></p>