Altered connectivity and network efficiency in preterm infants.

Abstract

Preterm birth is associated with altered maturation of large-scale functional brain networks during the neonatal period. Acquisition and preprocessing variability across neonatal resting-state fMRI (rs-fMRI) studies limits cross-site comparability and may obscure prematurity-related alterations. We combined rs-fMRI data from 90 local preterm infants scanned at term-equivalent age (TEA) with data from 160 infants in the Developing Human Connectome Project (100 term-born and 60 preterm) to evaluate cross-site harmonization and characterize edge-wise functional connectivity and network efficiency. Data were preprocessed using an identical pipeline in SPM and CONN toolbox, harmonized with neuroCombat, and analyzed using per-edge analysis of covariance for connectivity and linear mixed-effects models for graph efficiency. Harmonization substantially reduced site-related variance and improved cross-site comparability. Relative to term-born controls, preterm infants showed widespread edge-wise hypoconnectivity, with only limited focal hyperconnectivity. Graph-theoretical analysis revealed reduced local alongside modestly increased global network efficiency in preterm infants, whereas regional connection-class composition did not interact with group. These findings support the feasibility of harmonizing multi-site neonatal rs-fMRI data and suggest that prematurity at TEA is characterized by widespread functional hypoconnectivity and a shift in the balance between local network segregation and large-scale functional integration.