Growth faltering is associated with altered brain functional connectivity and cognitive outcomes in urban Bangladeshi children exposed to early adversity

Exposure to adverse conditions in early childhood has been shown to exert both proximal and distal effects on physical and psychological health and development. Stunting is regarded as a primary indicator of chronic malnutrition—a serious biological hazard with long-term developmental consequences—and affects a large number of children worldwide, especially in low-income countries. According to recent reports by UNICEF, World Health Organization (WHO), and the World Bank Group, 159 million children under age 5 years can be classified as stunted (i.e., having a standardized height-for-age [HAZ] score that is 2 standard deviations [SDs] below the median of the WHO reference). Chronic malnutrition resulting in stunting has previously been associated with delayed brain development and poor cognitive performance [1, 2], which in turn has a profound impact on the degree to which children can achieve their developmental potential [3]. Although it is well known that sufficient nutrients are necessary for brain and cognitive development [4], the mechanistic pathways by which malnutrition in early childhood relate to later cognitive outcomes remain unclear. This is, in part, attributable to a scarcity of research using both neuroimaging and sensitive behavioral measurement in low-income settings where growth faltering is most prevalent.

In the current study, we investigate associations between stunted growth as an indicator of chronic malnutrition and brain functioning in Bangladeshi children living in impoverished, low-resource neighborhoods. Specifically, we explore the association between HAZ and source-space EEG functional connectivity (FC), and whether brain FC mediates the relation between HAZ and cognitive functioning. EEG FC in different frequency bands has been shown to be a useful tool to investigate the development of the efficiency and organization of brain networks among typically developing children, as well as children exposed to early adversity [5‐9]. Variation in FC is often attributed to changes in the organization and functioning of brain networks. Abnormal patterns of FC within certain circuits due to biological adversity have been linked to deficits in later cognitive performance [10, 11]. For example, the FC in neonates’ subcortical, salience, and dorsal attention networks is correlated with maternal inflammation during pregnancy and predicts children’s working memory performance at 2 years of age [10]. It is therefore plausible that communication between cortical areas through neural oscillations in different frequency bands represents one pathway that is disrupted by chronic malnutrition during early childhood which, in turn, may lead to deficits in cognitive outcomes.

A large body of evidence from both human and animal studies supports the association between nutritional deficiencies (e.g., stunted growth) and atypical patterns of brain development [1]. For example, evidence from human postmortem studies revealed that 3- to 4-month-old infants suffering from malnutrition (indicated by low weight-for-age) showed reduced dendrite growth compared to well-nourished infants in the primary motor cortex (i.e., precentral gyrus) [12]. In addition, adults exposed to prenatal famine have white matter (WM) hyperintensities over the entire cerebrum on structural MRI. Increased WM volume among those exposed to famine may be caused by an inadequate supply of nutrients early in life to sustain and replace catabolized myelin and gliosis after myelin loss [13]. Animal models support histological evidence from malnourished infants in that studies of rodents have found that undernutrition is associated with reduced density of synapses and neurons in neural tissue over the cortex and alterations in callosal (interhemispheric) connections, likely caused by reduced neuron proliferation and changes in myelination and synaptic pruning [14‐16]. The neuronal and volumetric changes in the brain associated with undernutrition may lead to poor cognitive outcomes. A growing literature has demonstrated the adverse effects of stunting on children’s cognitive development that, in turn, is believed to contribute to worse educational and labor-market outcomes, including lower income and poorer productivity [17, 18]. In addition, faltered growth during infancy and childhood compared to later adolescence is more likely to cause negative long-term effects on adult health and capital [18, 19].

The first few postnatal years represent a period of rapid neural change [20] and a critical window during which experiences have strong effects on neural and cognitive development [21]. Children living in low-resource settings are often exposed to a variety of biological, psychosocial, and environmental adversities beginning early in life [22]. Given the critical knowledge gap concerning the neural pathways by which growth faltering in early childhood affects cognitive outcomes, it is important to examine the associations between HAZ, brain functioning, and cognitive outcomes in children living in low-resource environments with high rates of stunting.

The current study recruited two cohorts of infants and toddlers residing in an urban slum in Dhaka, Bangladesh. For the “infant cohort,” the length of the infants was measured at 3, 4.5, and 6 months, and resting-state (baseline) EEG data were collected at 6 months. For the older toddler cohort, the height of the toddlers was measured at 24, 30, and 36 months, and their EEG data were collected at 36 months. Brain FC between cortical regions was estimated after cortical source reconstruction of the scalp-level EEG, with the aim to reduce the effects of volume conduction on FC between electrodes [23]. We primarily focused on the global FC over the entire brain, while exploratory analyses were also conducted to examine whether FC within and between certain lobes (i.e., frontal, temporal, parietal, and occipital) varied as a function of adverse experiences. Brain FC in the theta, alpha, and beta frequency bands were examined because activation in these bands is commonly used to study the neural correlates of cognitive development in the area of attention, memory, and emotion processing in infants and young children [24]. We also explored connectivity in the gamma band since activity in this band has been associated with cognitive outcomes such as language development [25] and exposure to adversity in infancy [26]. Prospective cognitive outcomes in the current study were assessed at 27 months for the infant cohort and at 48 months for the toddler cohort.

Longitudinal path analysis was conducted to test the associations between HAZ, brain FC, and cognitive outcomes, and whether brain FC mediated the relation between HAZ and cognitive outcomes. In these analyses, socioeconomic status (SES) and family caregiving were included as covariates as they are associated with child development and may confound the relation between child growth and neurocognitive development [21, 27]. We additionally controlled for children’s head circumference because of previously shown associations with physical growth and cognitive development in children in low- and middle-income countries [28]. Given the global impact of undernutrition on brain functioning and anatomical development shown in the literature, we hypothesized that stunted growth would be prospectively associated with distinct patterns of brain FC between multiple regions and circuits in the brain (i.e., whole-brain or global FC). We further hypothesized that stunted growth would be prospectively associated with worse cognitive outcomes, and brain FC would mediate the link between child growth and cognitive functioning.

The current study provides the first evidence that physical growth early in life is associated with variations in network connectivity in the brain as inferred from theta- and beta-band FC between brain regions among children living in a low-income country. In a cohort of toddlers, we also show that brain FC is, in turn, prospectively associated with later cognitive performance and that there is an indirect effect of HAZ measured between 24 and 36 months on IQ measured at 48 months via FC at 36 months. In contrast, in a younger cohort of infants, no association was found between HAZ measured between 3 and 6 months and brain FC at 6 months or MSEL composite score at 27 months.

The associations between HAZ and whole-brain FC for the toddler cohort may reflect a broad deleterious effect of growth faltering on children’s brain functioning. The neural oscillations in the theta and beta bands have been associated with cognitive functions such as sustained attention, executive attention, and working memory [37, 41‐43]. Atypical patterns of FC in these frequency bands may therefore be associated with deficits in cognitive functioning in a number of critical domains. Interestingly, the pattern of associations between HAZ and FC in these two frequency bands was found to be slightly different, such that stunting was associated with stronger FC in the theta band for most of the brain regions, especially connections involving the frontal lobe, whereas the association between HAZ and FC in the beta band was more prominent for connections involving the occipital lobe. This discrepancy could be due to differences in the location of the cortical sources of theta and beta rhythms in early childhood.

One plausible explanation for the heightened FC among children with lower HAZ (including stunted children) is under/delayed synaptic pruning due to malnutrition, in turn leading to the failure to eliminate unnecessary connections as well as less organized and more superfluous pathways between brain networks [13, 44]. Children’s experience plays a key role in synaptic pruning, which begins in the first year after birth and continues through adolescence [45]. Delayed synaptic pruning may occur in stunted children due to a lack of stimulation and input from the environment. A second explanation of the negative association between HAZ and brain FC is that higher connectivity may reflect a more adaptive or compensatory neural response to make up for delayed brain anatomical development in growth faltered children. For instance, higher connectivity could indicate less efficiency of neural communications between cortical regions, which requires more effort for children who are stunted. This second explanation is consistent with findings from a recent fMRI study which demonstrated that maternal inflammation level (IL-6 concentration) is positively associated with functional connectivity in some brain networks and better developmental outcomes [46]. Indeed, the authors of that study suggested that increased FC may reflect an adaptive neurodevelopmental response to inflammation exposure. Finally, a third possibility is that greater functional connectivity across the brain may reflect less network segregation (differentiation) between functional submodules and a less mature connectivity pattern, which is more similar to the network organization found in younger infants [47]. Future research that makes use of sophisticated techniques such as graph theory will be useful to elucidate the likelihood of this hypothesis [6].

The current study also showed that global brain FC decreases from 6 to 36 months and the peak FC frequency increases with age (Fig. 1). This finding provides insights into the developmental course of EEG brain FC—in this case “phase-to-phase synchrony”—in early childhood. The increase of the peak FC frequency from 6 to 36 months is consistent with the known changes in EEG power in childhood, which could reflect typical neural maturational processes [48]. As noted above, the general decrease of brain FC with age may reflect increased synaptic pruning occurring in childhood. If this is the typical pattern of EEG FC over the course of childhood, higher FC among stunted children may suggest delayed brain development. Future research should continue to investigate whether changes in brain FC during childhood follow a linear or non-linear trajectory by including more age groups between 6 and 36 months and using different functional connectivity methods (see the results with “imaginary part of the coherency (iCOH)” in Additional file 1: Supplemental Information).

The absence of an association between HAZ and FC in the alpha band may be due to the fact that synchronization in this band was suppressed during the experiment when children were watching (attending to) a screensaver with abstract shapes, as alpha is known to be attenuated while engaging in attention-demanding tasks. A second possibility is that using a priori defined boundaries might overlook the most prominent alpha-band FC, as the individual alpha peak frequency changes with age over childhood (Fig. 1; also see [49]). Although additional analysis with FC defined with individual alpha peaks failed to show a significant association between alpha-band FC and HAZ (Additional file 1: Supplemental Information), future research may consider using individually defined boundaries to measure alpha-band power and FC in children.

A particularly compelling finding of the current study is that the association between child growth and developmental quotient (i.e., IQ) operated through brain FC in the theta and beta bands in the toddler cohort. Biological adversity related to inflammation has recently been shown to impact FC and cognitive abilities in studies using fMRI techniques [10, 46]. Our finding adds to this growing literature by showing that FC mediates the relation between biological adversity vis-à-vis growth stunting and later cognition, such that FC may serve as a neural pathway by which exposure to early adversity (including malnutrition) affects cognitive outcomes. As a practical matter, the current study suggests that EEG—a measure that is less expensive and easier to implement than MRI—has the sensitivity to detect differences in brain FC related to early adversity and later cognitive functioning in low-resource settings.

The high prevalence of stunting in the 36-month cohort highlights the severity of deficits in neurobiological development in this sample of Bangladeshi children. In this cohort, 33.1% of the children were classified as stunted. This prevalence rate is comparable to a report by the Bangladesh Demographic and Health Survey in 2014, which showed that the prevalence of stunting is 33.5% among children under age 15. In the current study, the prevalence of growth faltering at 6 months was relatively lower than that at 36 months, with approximately 16% of the infant cohort meeting criteria for stunting. The increased prevalence of stunting from 6 to 36 months is consistent with the standard trajectory of children with poor nutrition, such that children with poor early nutrition fall off the growth trajectories across early childhood as they age [50]. Moreover, consistent with the robust literature suggesting the role of SES in cognitive development [27, 51], the current study showed that SES was positively associated with IQ. Given a large number of variables that covary with child growth, researchers should be mindful that deficits in brain and cognitive development are likely caused by a constellation of interacting factors for children living in low-income countries [52].

The absence of an association between growth faltering and FC or cognitive outcomes in the 6-month infant cohort suggests that the impact of growth faltering on brain and cognitive development may build over time and either (a) not be detected in the first half year of life or (b) this association may not yet be present. This explanation is supported by a recent study reporting very small effect sizes in the association of growth measures during infancy and cognitive development among children living in low-income countries [28]. Alternatively, relatively lower variance in HAZ within the infant compared to toddler cohort (including fewer stunted children) may have reduced the ability to detect significant associations between HAZ and FC. Another possibility is that breastfeeding, which was prominent during the first 6 months, protected children in the infant cohort from growth faltering, or from the effects of growth faltering on brain and cognitive development [50]. The current findings for the infant cohort highlight the need for future research to follow these children longitudinally and monitor their growth trajectories throughout childhood.

The current study is the first to investigate the relation between growth faltering, brain FC, and cognitive outcomes in children living in a low-income country. Our findings indicate that growth faltering, an indicator of chronic malnutrition, is prospectively associated with exaggerated EEG functional connections in the brain, which in turn are prospectively associated with poorer cognitive outcomes in the first 5 years of life. Findings from the current study advance our understanding of the neural pathways by which faltered growth could be associated with cognitive development, and this advancement may have a substantial impact on developing efficient interventions for children living in low-income countries.