Spatiotemporal Development and Clinical Correlates of MRI-based Brain Myelination in Term- and Preterm-born Children.

Zhang, Yuqi; Li, Mingyang; Zhao, Ruoke; Liu, Tingting; Wu, Jiani; Xu, Xinyi; Chen, Ruike; Chen, Yiwei et al. · Radiology · 2025

prospective_cohort · Level II

Where this comes from

Abstract

Background The spatiotemporal progression of myelination and its associations with neurodevelopmental outcomes and perinatal risks remain insufficiently characterized in large-scale longitudinal studies. Purpose To characterize the myelination process across early childhood using the T1-weighted to T2-weighted MRI signal intensity ratio as a semiquantitative marker of myelination and elucidate its relationship with perinatal adversities and neurodevelopmental outcomes. Materials and Methods In this multicenter study, 3.0-T MRI data of term-born children from the Baby Connectome Project (BCP) (September 2016 to May 2020) and preterm-born infants collected locally at Zhejiang University (ZJU) (January 2019 to September 2020) were analyzed to extract the T1-weighted to T2-weighted signal intensity ratio. Statistical analyses included nonnegative matrix factorization to identify regions with distinct patterns of myelination (PMs), generalized additive mixed-effects models to characterize trajectories, and mediation and correlation analyses to assess brain-behavior relationships. Results In total, 412 children (mean age, 14.6 months ± 13.5 [SD]; 214 male; 307 from BCP and 105 from the ZJU local dataset) with 779 imaging sessions were included. Spatiotemporal maps revealed rapid myelination in the first 2 years of life. Seven PMs with unique developmental trajectories were identified. A full mediating effect of T1-weighted to T2-weighted signal intensity ratio was found on the age-related change of repetitive motor frequency (indirect effect β range, -0.40 to -0.24; false discovery rate [FDR]-corrected <i>P</i> < .001) and restricted behavior frequency (β range, -0.33 to -0.22; FDR-corrected <i>P</i> < .05). Compared with moderately preterm infants, extremely preterm infants had slower myelination across all PMs (<i>t</i> value range, 2.71-3.27; <i>P</i> < .01) and showed delayed fine motor skills (<i>t</i> = -2.47, [<i>P</i> = .03] at 4 months; <i>t</i> = -4.00 [<i>P</i> = .02] at 8 months). The spatiotemporal pattern of myelination in extremely preterm infants was less correlated with the normative term-born group (<i>r</i> = 0.807; <i>P</i> = .03) than that of moderately preterm infants (<i>r</i> = 0.952; <i>P</i> < .001). Conclusion The T1-weighted to T2-weighted signal intensity ratio from MRI was used to identify heterogeneous PMs and elucidated a critical link between myelination and neurodevelopmental outcomes in both term- and preterm-born children. © RSNA, 2025 <i>Supplemental material is available for this article.</i> See also the editorial by Widjaja in this issue.

Medical subject headings