In both voxel-wise GLM and tract-based linear regression analyses, preterm infants requiring C-section delivery had reduced WM microstructural integrity compared to those delivered vaginally, most prominently in the corpus callosum, internal capsule, and corona radiata. Notably, the difference between delivery mode of preterm infants was found after adjusting for gestational age at birth, gestational age at scan, 5-minute APGAR, birth weight Z-score, presence of preeclampsia, and presence of chorioamnionitis as covariates – and after excluding infants from multiple gestations and those with any maternal history of drug use. Although birthweight Z-score and presence of maternal preeclampsia differed significantly between infants delivered vaginally and via C-section, additional voxel-wise GLM analyses found no significant association between these risk factors and DTI metrics after controlling for delivery method and other clinical variables. In the sub-cohort excluding SGA infants, areas of significant association between delivery method and DTI metrics were limited compared to the primary cohort but still included pertinent WM tracts such as the corpus callosum, internal capsule, and corona radiata.
Our results provide the neurobiological indications and early WM microstructural correlates for clinical studies reporting that preterm infants who require C-section are at risk for adverse neurodevelopmental outcome. The difference in early WM microstructural development may be a consequence of multiple factors necessitating C-section delivery; nevertheless, our findings highlight the higher risks of neurodevelopmental delay among preterm infants delivered via C-section and the need for close follow-up and potential intervention to mitigate these risks in this vulnerable cohort.
The myelination of WM tracts as infants age reduces the random diffusion of water molecules and results in alignment of diffusion along WM tracts. Thus, changes in DTI metrics such as increased FA and reduced MD and RD are corollaries for WM maturation.(33) DTI metrics have also been shown to predict neurological outcome, with studies reporting associations between DTI metrics measured during term-equivalent scans and cognitive, motor, and visual outcomes in childhood and adolescence.(34, 35) Of note, some have reported associations between the development of cerebral palsy and other motor abnormalities and term equivalent DTI metrics in the same WM regions where we identified significant association between delivery method and DTI metrics such as the corpus callosum and internal capsule.(36, 37) Thus, term-equivalent DTI can potentially help identify infants at risk for adverse neurological outcomes with regards to their method of delivery, especially when neurological exam and standard neuroimaging might otherwise be unremarkable.
The relationship between delivery method and infant brain development has not been extensively studied. Deoni et. al have compared the DTI metrics between term infants delivered via C-section (n = 11) versus vaginally (n = 32).(25) Similar to our results, they reported higher FA values in WM tracts of term infants delivered vaginally compared to those delivered via C-section. We found higher MD and RD in addition to reduced FA in preterm infants delivered via C-section. Deoni et. al also noted slight variations in results with the inclusion of additional covariates, suggesting that factors beyond delivery method could be contributing to perceived differences in WM maturation. We attempted to address this by controlling for clinical variables known to be independently associated with WM maturation in infants (gestational age) in addition to those we found differed significantly between our two experimental groups (birthweight z-score and maternal preeclampsia), among others.(23) As prior reports have raised concern that exposure to general anesthesia, particularly inhaled agents such as sevoflurane, may contribute to delayed neurodevelopment and adverse long-term neurological outcomes, we also confirmed that exposure to general anesthesia during delivery was not independently associated with WM maturation in our cohort.(31, 32, 38, 39) Of note, Deoni et. al found no difference in WM tract DTI metrics when comparing separate cohorts of eight-year-old children delivered via C-section (n = 23) versus vaginally (n = 37). Thus, they proposed that WM microstructural differences in association with delivery mode may only be present in early (pre-discharge) MRIs and not resolve at older age – although they analyzed different children’s cohorts from infants.
As we demonstrate the relationship between delivery method and WM maturation on neuroimaging, the question of how delivery method may influence early brain development arises. It has been suggested that vaginal delivery is an important contributor to the normal development of infants’ gut microbiome and that alterations in this process (due to C-section) can disturb normal brain development.(40, 41) The impact of differential infant hormonal expression between vaginal and C-section deliveries on early brain development has also been proposed.(42) However, these explanations are complicated by studies reporting contrasting results regarding the association of delivery method with neurodevelopmental disorders.(19, 43, 44) Even among the numerous studies suggesting that delivery method may be associated with neurological outcomes, differences in methodology (such as additional clinical covariates that were controlled) and characteristics of experimental cohorts among prior reports complicate interpretation. Additionally, some investigators report that delivery method may be associated with neurological outcome only in the context of additional factors such as infant gender, APGAR scores, multiple gestations, and use of general anesthesia during C-section.(12, 16, 17, 32) Given the many maternal and infant factors that may influence need for C-section versus vaginal delivery, it is challenging to control for all covariates that may influence brain development independently or cooperatively with delivery method. Nevertheless, our study provides WM microstructural correlates for prior clinical studies reporting that preterm infants requiring C-section are at higher risk of neurodevelopmental delay. While our findings may be related to multifactorial circumstances necessitating C-section delivery, they highlight the need for close neurological follow up and timely interventions in these vulnerable children.
Exclusion of infants born SGA in the sub-cohort analyses limited the regions of apparent significant association between delivery method and DTI metrics when compared to primary cohort analyses. This suggests that the results of our primary analyses could be partially mediated by the lower birth weight and birth weight z-score among neonates born by C-section versus vaginally in the primary cohort. Indeed, prior reports have described the association between birth weight and DTI metrics as well as long-term neurological outcomes.(45–47) However, birth weight does not entirely explain the results of our primary analyses as evidenced by lack of significant independent association between birth-weight z-score and DTI metrics in primary and sub-cohort analyses as well as sustained significant associations between delivery method and DTI metrics in the corpus callosum, internal capsule, and corona radiata in our sub-cohort analyses.
Our study is limited by its retrospective, single center design and small sample size. Our small sample size also limits the power of our study in including multiple covariates, which may complicate the interpretations of our analyses on the independent associations of birthweight z-score, maternal preeclampsia, and general anesthesia use during delivery with DTI metrics. The small sample size especially limited the sub-cohort analysis after exclusion of 9 SGA preterm infants. DTI metric analysis can be influenced by acquisition protocol and timing, as well as processing methods used to generate DTI metric maps. Although we controlled for several demographic and clinical variables, including those that differed significantly between experimental groups, we could not account for all possible covariates that may influence delivery method and explain observed differences in DTI metrics between the experimental groups in this study. Of note, recent work has suggested that intrauterine inflammation impacts fetal and infant neurodevelopment and although we controlled for the presence of chorioamnionitis as a covariate, we did not collect placental pathology data to assess for other causes of intrauterine inflammation.(48, 49) We also described various indications for respective delivery method but could not control for all such indications given the widely varying delivery indications in our cohort. Finally, long-term neurological outcomes were not available in our dataset to corroborate clinical correlates of DTI findings.