Neuroimaging young children and associations with neurocognitive development in a South African birth cohort study

Abstract

Magnetic resonance imaging (MRI) is an indispensable tool for investigating brain development in young children and the neurobiological mechanisms underlying developmental risk and resilience. Sub-Saharan Africa has the highest proportion of children at risk of developmental delay worldwide, yet in this region there is very limited neuroimaging research focusing on the neurobiology of such impairment. Furthermore, paediatric MRI imaging is challenging in any setting due to motion sensitivity. Although sedation and anesthesia are routinely used in clinical practice to minimise movement in young children, this may not be ethical in the context of research. Our study aimed to investigate the feasibility of paediatric multimodal MRI at age 2–3 years without sedation, and to explore the relationship between cortical structure and neurocognitive development at this understudied age in a sub-Saharan African setting. A total of 239 children from the Drakenstein Child Health Study, a large observational South African birth cohort, were recruited for neuroimaging at 2–3 years of age. Scans were conducted during natural sleep utilising locally developed techniques. T1-MEMPRAGE and T2-weighted structural imaging, resting state functional MRI, diffusion tensor imaging and magnetic resonance spectroscopy sequences were included. Child neurodevelopment was assessed using the Bayley-III Scales of Infant and Toddler Development. Following 23 pilot scans, 216 children underwent scanning and T1-weighted images were obtained from 167/216 (77%) of children (median age 34.8 months). Furthermore, we found cortical surface area and thickness within frontal regions were associated with cognitive development, and in temporal and frontal regions with language development (beta coefficient ≥0.20). Overall, we demonstrate the feasibility of carrying out a neuroimaging study of young children during natural sleep in sub-Saharan Africa. Our findings indicate that dynamic morphological changes in heteromodal association regions are associated with cognitive and language development at this young age. These proof-of-concept analyses suggest similar links between the brain and cognition as prior literature from high income countries, enhancing understanding of the interplay between cortical structure and function during brain maturation.

A systematic review of MRI studies of language development from birth to 2 years of age

“…It is often observed also in MRI studies beyond those with a focus on language development that communities with higher educational level and SES seem to be more active to participate in the studies. There is a need for studies of early brain development and cognition in all socio‐cultural communities and especially from the countries of low income (Wedderburn et al., 2020). Considering these observations, we find it important for future studies to report adequate child and family demographics.…”

The impact of maternal nutrition on birth weight and brain development in children from low- and middle-income countries

“…In turn, poor maternal nutrition is one of several important determinants of adverse birth outcomes (19,34) such as LBW (26,51). Based on previous literature (58), we hypothesize that LBW will predict altered brain development and, consequently, poor neurodevelopmental outcomes (63). Poor birth and developmental outcomes may contribute to an intergenerational feedback loop (5).…”

“…The advent of neuroimaging has provided researchers with non-invasive tools to directly examine the effect of maternal nutrition on brain structure and function (Isaacs, 2013). Given the link between brain structure and neurocognitive outcomes (63,64), this provides an opportunity to further understand how maternal nutrition and LBW may influence later infant outcomes. It is recognized that the prevalence of food insecurity, malnutrition and LBW is higher in LMICs ( 35), yet most neuroimaging research thus far has been done in high-income countries (65).…”

Feasibility of FreeSurfer processing for T1-weighted brain images of 5-year-olds: semiautomated protocol of FinnBrain Neuroimaging Lab

“…In this article, we propose a dichotomous rating scale for inclusion and exclusion of images, combined with a post-processing quality control protocol to visually confirm high quality data on a ROI level. For the automated segmentation tool in this protocol, we chose FreeSurfer based on the following practical advantages: 1) FreeSurfer has been validated for use in children between ages four and eleven years (Ghosh et al, 2010), and multiple studies have used FreeSurfer to find brain associations between brain structure and risk factors or cognitive differences in children (Black et al, 2012;Clark et al, 2014;Wedderburn et al, 2020); 2) FreeSurfer provides a method to accurately assess image quality and to fix certain types of errors via Freeview; and 3) Rigorous quality control protocols, such as the one provided by the ENIGMA consortium (Enhancing Neuro Imaging Genetics through Meta Analysis; http://enigma.ini.usc.edu/), already exist for FreeSurfer to make final quality assessment on such a level that allows the researchers to exclude single ROIs with imperfect segmentation. We decided to use the ENIGMA quality control protocol, as it is widely used and accepted (Thompson et al, 2020), and has been successfully implemented for both adults (Thompson et al, 2020) and children (Boedhoe et al, 2018;Hoogman et al, 2019).…”

“…FreeSurfer (http://surfer.nmr.mgh.harvard.edu/) is an open source software suite for processing brain MRI images that is commonly used in pediatric neuroimaging (Al Harrach et al, 2019; Barnes-Davis et al, 2020; Black et al, 2012; Boutzoukas et al, 2020; Clark et al, 2014; El Marroun et al, 2016; Garnett et al, 2018; Ghosh et al, 2010; Lee et al, 2017; Nwosu et al, 2018; Phan, Smeets, et al, 2018; Ranger et al, 2013; Roos et al, 2014; Wedderburn et al, 2020). The automated FreeSurfer segmentation protocol utilizes surface-based parcellation of cortical regions based on cortical folding patterns and a priori knowledge of anatomical structures (further technical information in Dale et al, 1999; Fischl, Sereno, & Dale, 1999).…”