To the best of our knowledge, this is the first study to address the association between delayed white matter maturation and NDI in preterm infants with NEC and without apparent brain abnormalities. Compared with preterm infants without NEC, the NEC group had a significant NDI at 18 months of CA. DTI analysis showed that the NEC group had increased MD of sCC and ltCST, indicating delayed microstructural maturation at TEA. Moreover, MD of sCC was negatively associated with BSID-III language and motor composite scores.
We found that the NEC group had smaller white and deep gray matter size and decreased total brain volume although the associations were not statistically significant. We assume that impaired brain volume in preterm infants with NEC in previous studies is due to confounders, including total intracranial volume or head circumference [20, 21]. Although the associations between NEC and brain volume were not evident, the results implied that brain volume reduction in the NEC group was due to delayed brain growth rather than secondary atrophy or brain injury caused by NEC.
In developmental assessment, NEC was a significant risk factor of NDI only in motor function. Considering our study set cut-off value of NDI with composite score under − 2 standard deviation, our result might have reflected moderate to severe NDI. As NEC group had significantly poor BSID composite scores in every index, further study with large sample size would be needed to elucidate its relationship. This study showed that preterm infants with NEC exhibited white matter maturation delay as network parameter distinctions in DTI analysis. Maturation delay of sCC and ltCST with prolonged MD was prominent, reflecting that neuronal fibers were loosely connected and thereby poorly integrated in the NEC group. This implies that NEC plays a substantial role in white matter maturation, along with prematurity, contributing to NDI. The corpus callosum (CC) is the largest white matter bundle that conducts inter-hemispheric information. In particular, sCC is the most integrated lesion connecting the temporal, parietal, and occipital cortices. It plays a major role in the transhemispheric processing of visual and acoustic data and is myelinated at 3–4 months of age [22, 23]. In previous DTI studies, delayed maturation of CC was implicated in cognitive function [22, 24], gait, and motor coordination  of preterm infants. Our study is in line with those studies, emphasizing that maturation of CC is associated with NDI in preterm infants with NEC. Moreover, we found that MD of sCC could be a biomarker of later NDI, especially in cognitive and language functions, reflecting its correlation with BSID-III composite scores.
Interestingly, an increase in MD was not followed by a decrease in the FA in the CC. Although both FA and MD are primarily calculated parameters that reflect the degree of myelination, they can contain different characteristics. FA measures the degree of anisotropy within a voxel, implying the direction of myelination, whereas MD measures the average degree of water molecule diffusion . CC is myelinated at 3–4 months of age, suggesting less anisotropy and less myelination at TEA in this study. Thus, we assumed that delayed maturation of the NEC group would manifest with an increase in MD as early as TEA rather than a decrease in FA . Given that altered brain development underpins maturation-dependent vulnerability, the insignificant difference between the two groups in the myelination of gCC could be explained by the general myelination pattern in the posterior-to-anterior direction .
The observed association between white matter maturation and NEC may reflect multifactorial etiologic factors of NEC, including intestinal immaturity, hypoxic-ischemic injury, infection, and inflammation. In particular, inflammation has been hypothesized to be the principal cause of NDI. Previous studies have shown that inflammatory conditions accompanied by sepsis are associated with altered brain microstructural perturbing white matter microstructural integrity [27, 28]. Moreover, Alshaikh et al.  showed that inflammatory conditions played a crucial role in white matter abnormalities, even in the absence of evident brain injury in a meta-analysis. Let alone the abnormalities in the infant period, Dubner et al.  showed that inflammatory conditions in the neonatal period had altered white matter microstructure in CC with delayed cognitive function at 6 years of age. As NEC has been implicated in the pathogenesis of infection and inflammation in one of the neonatal diseases, our results are in line with previous reports emphasizing the role of inflammation in NDI of infants with NEC. The mechanism of white matter delayed maturation in NEC is not well understood. However, multiple studies have reported elevated levels of inflammatory cytokines and disruption of the blood-brain barrier in animal models of NEC [31, 32]. We assume that the systemic inflammatory response causes widespread oxidative stress in the brain as NEC progresses. Subsequently, pre-oligodendrocytes are injured, which has marked vulnerability to cytokine injury and reactive oxygen, in response to the activation of astrocytes and microglia. Injured pre-oligodendrocytes may have failed to evolve into myelin-producing oligodendrocytes, thereby causing delayed white matter maturation and NDI . Moreover, conditions followed by NEC including acidemia, sepsis and management followed by NEC including mechanical ventilation, systemic antibiotics and indwelling catheters may have worsened systemic inflammation and disrupted gut microbiota for early brain development.
Growing evidence suggests that early postnatal growth and nutrition during the NICU period affect brain volume expansion and white matter maturation [4, 34, 35]. These findings suggest that nutrition and growth are possible confounding factors of NDI in infants with NEC. To reflect their confounding effects, we analyzed white matter maturation adjusted for duration of TPN days. Remained significance, after adjustment for duration of TPN days, implies NEC is an independent risk factor for NDI; however, malnutrition may simultaneously increase the risk of worsening neurodevelopmental outcome.
This study had several limitations. One major limitation is the small sample size in the NEC group, which may lead to subject heterogeneity and skewed interpretation of the data. Greater statistical power could be gained by more balanced sample size or multicenter imaging data. Second, due to poor quality images with motion artifacts, a substantial amount of data was discarded during image processing; thus, caution must be exercised in the interpretation of results. However, in this study, all scans were performed during natural sleep, and all infants were carefully monitored with pulse oximetry and supervised by a skilled physician without sedation to ensure safety during the MRI scanning procedure. Compared with images of adults, there are many methodological challenges in obtaining neonatal MRI images. Finally, important questions remain regarding the contribution of genetic and postnatal environmental factors to the link between DTI findings and NDI. Although we corrected for important confounders such as GA and TPN days of nutrition, variables reflecting inflammatory status, including culture-proven sepsis and BPD, might have affected NDI.