Pre- and postnatal brain development is a complex process that is under active investigation, and it is a continuous and dynamic process that extends well into early adulthood (Stiles, 2008). Advances in magnetic resonance imaging (MRI) have provided researchers with tools to investigate developmental changes of cerebral gray and white matter. Diffusion tensor imaging (DTI) has been shown to capture white matter development, which is intrinsically linked to changes in the brain’s efficiency and related cognitive, behavioral, and psychopathological outcomes (Mabbott et al., 2006; Nagy et al., 2004; Filley and Fields, 2016). DTI characterizes white matter microstructure by measuring the orientation and integrity of white matter tracts (Mukherjee et al., 2008). This technique enables in vivo examination of tissue microstructure and provides quantifiable metrics relevant to both brain development and injury (Pecheva et al., 2018). Metrics obtained from DTI are highly sensitive to developmental shifts and white matter changes associated with prematurity and perinatal risk factors (Anjari et al., 2009; Ball et al., 2010; Chau et al., 2009; Zwicker et al., 2013). White matter pathways play a central role in the emergence of cognitive and behavioral functions, and their maturation follows a protracted, region-specific developmental course (Imperati et al., 2011; Lebel and Deoni, 2018). Building upon this, Nivins et al. (2023) studied how neonatal non-imaging measures (birth weight [BW], gestational age [GA], and head circumference at birth [HC]) were associated with imaging-derived measures of brain volume and white matter microstructure at school age (9–10 years). Notably, only birth weight showed some predictive value for white matter microstructure via the tract-based spatial statistics (TBSS) analysis in that study. In contrast, BW, GA, and HC are well-established specific indicators of fetal growth and early brain development, and prior studies have shown their associations with later white matter organization and neurodevelopmental outcomes (Jin et al., 2019; Solis-Urra et al., 2022). Furthermore, studying children aged 8–10 years is particularly informative, as this developmental period is characterized by relatively stable white matter maturation prior to pubertal changes, thereby reducing additional developmental variability (Lebel and Deoni, 2018; Simmonds et al., 2014). Methodologically, regional or white matter skeleton-based approaches often lack the sensitivity and spatial precision needed to detect tract-specific microstructural alterations, whereas fiber-tract–based analyses provide greater anatomical specificity and improved sensitivity (Short et al., 2022). Moreover, whereas prior studies such as Nivins et al. (2023) primarily focused on term-born children, our inclusion of participants spanning the full gestational age spectrum enables a more comprehensive evaluation of perinatal influences across a broader developmental range. Motivated by these limitations, the goal of the current study was to extend previous findings by applying a more spatially sensitive fiber tract-based DTI analysis framework (Verde et al., 2014) in a cohort with broader variability in perinatal factors, allowing comparison with results reported by Nivins et al. (2023).

Here, we aimed to investigate how common neonatal indicators of health (BW, GA, and HC) and brain development are related to DTI-based measurements of white matter microstructure in children aged 8–10 years.

Of the 148 participants, 31 were excluded during the quality control process, leaving 117 participants for analysis. Their baseline demographic and perinatal characteristics are summarized in Table 1. There were 94 participants scanned with the three-shell diffusion MRI sequence and 23 participants scanned with the single-shell diffusion MRI sequence. Gestational age was categorized following WHO guidelines as full term (37 ≤ GA < 42 weeks), moderate to late preterm (32 ≤ GA < 37 weeks), very preterm (28 ≤ GA < 32 weeks), and extremely preterm (GA < 28 weeks).

In this study, we found evidence suggesting that neonatal health indicators, particularly gestational age (GA) and birth weight (BW), are associated with white matter microstructure at ages 8–10 years as measured by DTI. While previous studies have shown associations between perinatal factors and brain volumes in this age group (El Marroun et al., 2020; Davis et al., 2011; de Kieviet et al., 2012; Catena et al., 2019; Cheong et al., 2008), our findings extend this understanding to white matter organization. Notably, GA and BW showed a widespread pattern of positive associations with FA and AD, whereas HC demonstrated only limited relationships, suggesting that early growth measures may differentially influence later white matter development. These findings highlight the lasting impact of early-life biological status on white matter maturation and underscore the importance of considering perinatal factors when interpreting later neurodevelopmental outcomes.

Our findings both align with and differ from prior research in important ways. While earlier DTI studies using TBSS found limited associations between perinatal factors and diffusion metrics (Nivins et al., 2023; Ou et al., 2017), our fiber-tract analysis framework revealed broader association patterns. TBSS is well-known to serve as a hypothesis-generation tool due to its whole-brain, white matter skeleton-based analysis framework (Smith et al., 2006; Bach et al., 2014), but it is also known to be less sensitive than fiber-tract analysis frameworks such as ours or Automated Fiber Quantification (AFQ) (Bach et al., 2014; Yeatman et al., 2012; Kuchling et al., 2018; Wang et al., 2022). This may reflect the greater sensitivity of the fiber-tract-based analysis in detecting associations between perinatal factors and white matter microstructure, compared to previous TBSS-based studies (Nivins et al., 2023; Ou et al., 2017).

In contrast to prior studies showing strong associations between birth HC and later brain volume measures (Catena et al., 2019; Cheong et al., 2008), we found that HC was associated with only a small number of white matter tracts in our tract-based DTI analysis. This pattern suggests that HC may be more closely related to global brain size than to tract-specific microstructural organization, consistent with the view that volumetric and diffusion-based microstructural measures reflect distinct aspects of neurodevelopment.

Our sample includes a broader range of gestational ages, from extremely preterm to full term, which may help explain the widespread associations with GA observed in our results, in contrast to studies such as Nivins et al. (2023) that focused primarily on term-born children. Importantly, among the 62 preterm participants, the majority were moderate or late preterm (16 moderate and 37 late), with a small number of very (n = 7) or extremely preterm children (n = 2). This distribution indicates that our findings are unlikely to be driven by a highly premature subgroup. Including children born preterm allowed us to assess associations across a broader developmental spectrum and to capture white matter variability related to lower GA. This interpretation aligns with previous literature showing that preterm birth—particularly at lower gestational ages—is associated with reduced FA, elevated RD, and delayed microstructural maturation (Kelly et al., 2016; Thompson et al., 2014; Young et al., 2018). Accordingly, we focus on the broader pattern of associations across the white matter network rather than interpreting each tract individually. The widespread involvement of major projection and association pathways suggests that perinatal factors may exert diffuse influences on the development of long-range connectivity, consistent with the global nature of early growth-related risks.

In the post hoc analysis, AD showed the greatest number of significant tract-level associations with perinatal factors compared to FA and RD, although the magnitude of these effects was generally small. Although the most rapid phases of axonal growth occur earlier in development, AD remains sensitive to residual variation in axonal organization that persists into late childhood (Genc et al., 2023; Dimond et al., 2020). This continued sensitivity likely explains why neonatal factors showed stronger and more widespread associations with AD than with RD, which is more specifically influenced by myelination processes (Song et al., 2005). Taken together, these findings suggest that higher GA and BW–and, to a lesser extent, HC–are associated with more advanced white matter organization in school-age children.

Several limitations warrant consideration. First, our moderate sample size (n = 117) and use of a single 3 T MRI scanner may limit the generalizability of our findings. Second, although the inclusion of children ranging from extremely preterm to full term allowed us to examine perinatal influences across a broad developmental spectrum, this heterogeneity may also introduce confounding effects related to differing medical and developmental trajectories. Third, neurodevelopment is shaped by complex interactions among genetic, environmental, and postnatal influences, none of which could be fully captured in the present analysis. In addition, although we included twin status as a covariate, within-pair relatedness was not modeled, which may have introduced non-independence in the data. Future work should incorporate analytic approaches that account for familial clustering (e.g., mixed-effects or GEE models). Finally, we did not apply formal multiple-comparison correction across tracts. Although white matter pathways are highly correlated and our analyses were investigative in nature—conditions under which traditional correction procedures may be overly conservative—this remains a limitation, and the findings should be interpreted with appropriate caution. Accordingly, the present results should be viewed as hypothesis-generating rather than as definitive evidence of tract-specific effects.

Future directions include expanding the analysis to the full EBDS cohort (over 300 participants scanned at age 8–10), which will allow for validating the current findings in a larger, more diverse sample once ongoing data harmonization is complete. Because these scans were acquired across three different 3 T Siemens MRI scanners, additional harmonization will be required to ensure consistency across scanners. In addition, the study plans to implement a more detailed analysis of white matter microstructure using the Neurite Orientation Dispersion and Density Imaging (NODDI) model, which will allow for a more refined and biologically specific understanding of white matter development in childhood.

In conclusion, we found that BW and GA—more so than HC—were associated with widespread patterns of differences in white matter organization in school-age children. These associations were most consistent for AD, suggesting sensitivity to long-term variation in axonal properties. Overall, our findings suggest that early-life biological factors may leave a measurable imprint on later white matter development.