Introduction
Diet is an important modifiable factor that modulates brain development by supplying energy and nutrients. Previous studies have established that poor diet quality during the fetal and early postnatal period impacts neurodevelopment in early life due to the increased risk of essential nutrient deficiency or protein-energy malnutrition [
1‐
3]. Beyond nutrition in early life, children develop distinct dietary patterns as they grow. In the meantime, structural and functional brain plasticity continues to develop throughout childhood and adolescence [
4]. Thus, the brain remain vulnerable to poor nutrition during this period of ongoing growth and modeling.
As a complement to the evidence regarding nutrients and neurodevelopment, there has been increasing interest to the relationship between overall dietary patterns and children's brain development. Dietary patterns, which can be quantified by predefined indices (e.g., diet quality score) or derived empirically from dietary data, are closely related to eating behaviors in the real-world setting by considering how foods and nutrients are combined [
5], and presenting the ability to detect the variation of dietary composition and its effect on health. Examples that are often studied are: prudent or healthy dietary patterns, often featured as high whole grains, fruits and vegetables intake; western-like or unhealthy dietary patterns, characterized by high intakes of refined grains, saturated fat, and sugar. Recent studies have shown that dietary patterns are established in early childhood and remain relatively stable into adulthood, with less healthy patterns showing the strongest stability [
6‐
8]. With ongoing global nutrition transitioning towards unhealthy diets [
9], more than half of children have poor-quality diets in low-, middle-, and high-income countries [
10‐
13], raising concerns about its cascading effect on children’s brain development.
A growing body of research has identified consistent temporal associations between children’s dietary patterns and cognitive performance in cohort studies. For instance, a better diet quality [
14] or a prudent dietary pattern at different ages through childhood is linked to higher intelligence quotient (IQ) scores of children and adolescents [
15], even when accounting for socioeconomic status and home environment. In contrast, children and adolescents with higher adherence to a western-like dietary pattern showed diminished cognitive functioning and lower IQ scores [
16,
17]. Among the complex mechanisms underlying the relation between diet and cognitive performance, structural brain alteration may be a detectable neurobiological marker on the pathway. Accumulating evidence from observational studies suggests that a Mediterranean diet, which shows neurocognitive protective effects, is associated with global brain morphological alteration in adults, including total brain volume, gray and white matter volume, as well as cortical thickness [
18]. Although studies among children are sparse, neuroimaging studies have identified positive links between IQ and alterations in both global brain volumetric measures in children using structural magnetic resonance imaging (MRI) [
19,
20]. Together with evidence on the effect of several individual micronutrients [
21] and macronutrients [
22] on the brain, global brain morphological changes may also underlie the associations between dietary patterns and cognition in children. However, there is a lack of knowledge as to whether dietary patterns in children are associated with global brain morphology and whether brain morphology underlies the associations between dietary patterns and cognitive performance.
Dietary effects on neurodevelopment may also be specific to certain brain regions. The hippocampus and amygdala may be particularly sensitive to the effects of specific dietary patterns, such as high-fat diet diets. The hippocampus is involved in integrating multimodal information and is critical for memory, learning, and is also involved in appetitive, digestive and learned eating behaviors [
23]. The amygdala has been implicated in feeding, the reward systems associated with eating, and modulating food consumption [
24]. Animal studies have found that a western diet has adverse effects on hippocampal dependent memory function and neurogenesis by triggering metabolic alterations [
25]. Additionally, western diets also induce inflammation in the hippocampus [
26,
27], and enhance amygdala-dependent, cue-based memory in juvenile mice [
28]. As a result of adapting to diet-induced endocrine abnormalities, structural alteration may occur in the hippocampus and amygdala. Moreover, structural changes may also be a neurobiological marker of functional disruption. A longitudinal study of seniors has found that individuals with higher adherence to a Western-style diet had decreased left hippocampal volumes four years later [
29]. It is, however, unclear whether different dietary patterns are associated with the volumetric differences of the hippocampus and amygdala in children. Only one recent study in children investigated the effect of a western diet on hippocampal and amygdala volume, but did not observe an association. However, they found that increased fat consumption was related to a smaller left hippocampal volume [
30]. This study has a small sample size and the western diet was calculated by only combining percentage of daily calories from fat and sugar, which may not capture the Western dietary pattern consumed in the general population.
Within this context, we aimed to examine the prospective association of dietary patterns in infancy and mid-childhood with brain morphology assessed at the age of 10 years in a large population-based cohort, and to examine whether dietary pattern-related differences in brain morphology mediate associations of dietary pattern adherence with full scale IQ at age 13 years. Based on evidence from animal studies, we examined the relationship between dietary patterns and two regional brain volumes (hippocampus and amygdala). Due to the paucity of information on the association of overall diet with brain morphology in the literature, we used an exploratory approach that involves global volumetric measures (total brain, cerebral white matter, cerebral gray matter), and subcortical volumetric measures (hippocampus and amygdala). In addition, we studied surface-based brain measures, which provide a different approach to evaluate the overall cortex, serving as exploratory analyses of whole brain measurements of gyrification, surface area and cortical thickness. We used two approaches to define dietary patterns: a priori-defined diet quality scores based on national dietary guidelines, and a posteriori-derived dietary patterns using principal components in the population. For global and subcortical volumetric measures, we hypothesized that higher adherence to a western-like dietary pattern and lower diet quality score would be associated with smaller global brain volumes at 10 years of age, and would be associated with smaller hippocampal and amygdala volume. We expected that positive associations would be found for a prudent dietary pattern or a better diet quality. For surface-based measures, we did not have prior hypotheses. Moreover, we expected that differences in brain morphology in relation to dietary pattern adherence would mediate the relationship between dietary pattern adherence and full-scale IQ. The findings of this study will facilitate future hypotheses on overall diet and brain development, as this has yet to be studied in children in a population-based setting.
Discussion
This large population-based prospective cohort study explored associations of different dietary patterns in infancy and in mid-childhood with global, regional brain volumes and surface-based brain morphometry when children were 10 years of age. Diet quality at one and eight years-of-age as assessed by adherence to dietary guidelines was low to moderate. Global brain volumes at age 10 were negatively associated with dietary patterns characterized by high intake of snack and processed foods at age one and age eight years. Interestingly, global brain volumes were positively associated with a dietary pattern characterized by whole grains, soft fats and dairy at age eight years. No volumetric differences were found in the hippocampus or amygdala for dietary patterns at one and eight years-of-age. Higher DQS-8y or higher adherence to the aforementioned pattern with high intake in whole grains, soft fats and dairy at age eight years was associated with a greater gyrification and larger surface area in widespread areas of brain, with significant areas primarily clustered in association cortices, notably in the prefrontal cortex. Furthermore, global brain volumes, gyrification and surface area mediated the relationships between dietary patterns at ages one and eight years and children’s IQ at the age of 13 years. It is important to note that our study can only be interpreted in light of dietary patterns as a whole, but not on single food groups. In the following discussion, we compared the dietary patterns high in snacks and processed foods intake with western-like dietary patterns, while the dietary patterns high in whole grains intake were compared with prudent dietary patterns due to their high factor loadings for those food groups. Overall, our findings suggest that having a prudent dietary pattern in school age, specifically one rich in whole grains, soft fat and dairy, is linked to larger global brain volumes, whereas consuming a western-like dietary pattern in infancy and school age is associated with lower global brain volume. Our study provides novel information that extend earlier studies examining the association of human overall diet with brain health in childhood.
Our findings of the association between global brain morphology and children’s dietary patterns in early- and mid-childhood, especially those related to a prudent dietary pattern and a western-like dietary pattern, complement previous studies showing that overall dietary patterns are associated with school attainment [
52,
53] and cognitive performance [
15‐
17,
54,
55] in children. Considering that total brain volume [
56], cerebral gray matter volume [
57] and white matter microstructure [
58] have been associated with cognitive ability of children, our results suggest that neuroanatomical correlates may underlie the association between dietary patterns and cognitive development in children. This hypothesis was further supported by the mediation analyses in this study, implying a mediating role of brain morphology in the association of dietary pattern and cognitive performance in children. Furthermore, our results indicate that a prudent or better diet quality was associated with brain regions functionally implicated in appetite regulation. Among those regions, most clustered in the dorsolateral prefrontal cortex (DLPFC). The prefrontal cortex is a complex region, but is notably involved in dopamine-mediated executive function, regulation of reward, and the inhibition of impulsive behaviors. Specifically, the DLPFC is associated with satiety, food craving, and executive functioning [
59,
60]. A randomized controlled trail (RCT) in healthy young men with normal BMI found increased neuronal activity in right DLPFC reduced overall caloric intake and diminished self-reported appetite scores [
61], which may suggest DLPFC involvement in food intake-related control mechanisms. However, it is unclear to what extent the functional connectivity translates into differences in brain morphology, although theories of gyrification support a relationship between folding patterns of the brain and increased connectivity [
62]. Caution is needed regarding the translation of functional connectivity studies with the observed association between dietary patterns and brain morphology.
Among all significant associations, a prudent dietary pattern at age eight years was positively associated and a western-like dietary pattern at age one year was negatively associated with global brain volumes at age 10 years after multiple testing correction. The association of ‘Snacks, processed foods and sugar’ dietary pattern at one year with cerebral white matter volumes persisted after controlling for intracranial volume, suggesting that regional cerebral white matter development may be specifically susceptible to a western-like diet specifically in infancy. A possible mechanistic interpretation is that overall diet influences brain development if adherence to a dietary pattern with a higher risk of nutrient inadequacy occurs during a period of high need, considering the rapid rate of brain development within early life [
1,
63]. The majority of white matter myelination occurs during the first 2 years of life [
64]. Meanwhile, research has shown that children with a western-like dietary pattern were exposed to excess fat and sugar intake or inadequate nutrient intake, such as omega-3 polyunsaturated fatty acids [
65], which serves as an important nutrient for nerve cell myelination and has shown beneficial effect on cognitive impairment [
66,
67].
Contrary to the animal studies, our findings did not support our hypothesis that overall diet is associated with smaller hippocampal and amygdala volumes. We are aware of only one study investigating the effects of a western diet on hippocampal and amygdala volume in five- to nine-year-old-children. Stadterman et al. [
30] reported a link between percentage of daily calories from fat, but not western diet, with an isolated decreased left hippocampal volume, but no relationship was found with the right hippocampal or amygdala volume. Their study calculated western diet as a summed percentage of daily calories from fat and sugar and the sample size was small (n = 21), however, it represents a promising step towards understanding the impact of western-like diet on hippocampus and amygdala development in children. We build upon their study investigating the association between a western-like diet and the left hippocampus but found no association. Our large sample size and the population-based sample, coupled with our approach to derive western-like dietary patterns, likely better resembles actual eating habits, although we did not observe evidence for any effects of either total or left hippocampal volume. Future research with longer follow-up time and in different study population is needed to ascertain the association of hippocampus and the amygdala with western-like diet in children.
Although the etiology of the relationships between dietary patterns and neurodevelopment remains unclear, there have been potential mechanisms proposed. First, the effect of dietary patterns on brain morphology may be mediated by epigenetic mechanisms. Diet, as an epigenetic programming regulator, affects multiple genes expression at levels of transcription, translation and post-translational modification. Subsequently, the variation in gene expression directly regulated by nutrition influences several neurobiological processes, including neurogenesis, synaptic plasticity and neuronal connectivity [
68,
69], which may lead to rearrangement of brain structure. The epigenetic mechanism can be supported by a study using data from the Barbados Nutrition Study, which followed a group of adults who had been hospitalized during their infancy for protein-energy malnutrition. Compared with a control group of adults who were not exposed to malnutrition in their childhood, the researchers identified 134 nutrition-sensitive, differentially methylated genomic regions using epigenome-wide analysis, and found that methylation at some of these sites were associated with cognitive outcomes (i.e., IQ and attention) [
70]. Second, differences in dietary patterns can induce differences in metabolic changes underlying brain morphological development. For example, brain-derived neurotrophic factor (BDNF), which plays a pivotal role in glucose and energy metabolism, is related to the morphological variation of the hippocampus and prefrontal cortex [
71], and can be influenced by diet. High-fat diets have been found to increase oxidative stress, inflammation and further interfere with the level of BDNF [
72]. Regarding healthy dietary patterns, a randomized control trial study in adults found a higher plasma BDNF levels in an experimental group with a Mediterranean diet intervention compared to a control group [
73]. Furthermore, decreased BDNF has been linked to cognitive decline [
72]. While these studies were performed in adults, there may be expectations of even greater differences in children, considering the high energy consumption associated with neurodevelopment. Further, the evidence showing dietary patterns were associated with cognitive performance in children, suggests shared brain metabolic pathways that underlie brain morphological changes in children and adults. Last, diet could affect the gut microbiota which in return alter the host’s physiological responses and associated structural adaption. Short-chain fatty acids (SCFAs) are the most studied metabolites produced by microbes. Foods high in dietary fiber, such as fruits, vegetables and whole grains, increase the levels of SCFAs through gut microbial fermentation [
74]. Results from animal studies indicate that SCFAs might influence gut-brain communication and brain function directly or indirectly through neurochemical pathways [
75].
By reporting associations of dietary patterns in early and mid-childhood with global and regional brain volumes, our findings underline the need for nutrition research related to child brain development, notably in middle childhood. Nutrition has been hypothesized as an aspect of the experience-dependent environment that can influence neurodevelopment [
76]. However, most studies on nutrition and brain development have focused on early life nutrition, while few studies have examined children in the ‘forgotten years’. In fact, dietary patterns change drastically from milk-based during infancy to omnivore patterns in childhood. Previous work from our group showed that diet quality at one year-of-age changes considerably by eight years-of-age [
12], indicating that relative stability of dietary patterns may occur after the age of one year. Another longitudinal study including five European countries suggests that dietary patterns are established between one and two years and remain stable to eight years of age [
8]. In addition, brain metabolism associated with neurodevelopment is high, with a considerable need of energy. This changes at around the age of four to five years, coincident with the slower rate of growth during that age. Overall, changes in dietary patterns across childhood, highlight the importance of investigating the effect of diet not only in infancy, but also in mid-childhood.
Most dietary interventions for brain development to date have targeted micronutrient supplementation, showing positive effects of nutritional supplementation in certain cognitive domains in nutrient-deficient children [
77]. Some interventions which used food supplementation (e.g., fortified foods) in early childhood [
1] or in children with atypical neurodevelopment [
78] have shown short-term effects on cognitive and motor development. Evidence from those RCTs is intriguing and denotes a potential causal link between nutrition and brain development. However, some RCT studies that provide supplements of a single nutrient or food in certain groups of children failed to detect changes in cognitive performance [
79,
80]. It is noteworthy that none of RCTs considers overall diet at the baseline in their study design, which may partly explain the inconsistency in findings. This could be explained by an adapted hypothetical scenario in which the effects of nutrient adequacy and overall diet quality may show an interacting effect on children’s cognitive development [
1], thus reducing the effect of nutrient supplementation. Moreover, it is largely unknown whether the effect of nutrition on brain development could be detected by objective anatomical measurement, like brain volumetric alteration. We filled in the gap of lack of evidence in a healthy pediatric population with a long follow-up period.
The strengths of our study are the population-based prospective design, availability of multiple covariates, and large-scale neuroimaging in children. In addition, we defined dietary patterns using a priori diet quality score and a posteriori dietary patterns based on food groups, which captured unrelated eating patterns in the population. Therefore, the results of our study provide a public health message on the relationships of dietary patterns and brain development in children.
The main limitation is the design of the study. Although longitudinal, the design is in essence cross-sectional, as we lack repeated measures of diet and brain imaging. Thus, our results cannot infer causality between dietary patterns and brain morphometry. Future studies with repeated measurements of dietary patterns and brain morphology, preferably those with an intervention component, are necessary to better understand the direction of the association. Another limitation involves assumptions associated with calculating the a priori and a posteriori dietary patterns analysis, which can influence the interpretation and limit the comparability of the study with others. Such assumptions include how food items were clustered into food groups, the number of principal components to be retained. Additionally, dietary patterns are likely to be a part of lifestyle behaviors, and thus there is the potential for residual confounding. However, after adjusting for BMI at the age of 10 years in the model, dietary patterns were still significantly associated with brain morphology. Several other limitations should also be acknowledged. First, the use of FFQs to quantify dietary intake is subject to measurement error [
81]. However, the FFQs used in our study were validated against three 24 h recalls in Dutch children at age one year and against the doubly labelled water method in Dutch children at age eight years, and had sufficient capacity of ranking participants with regard to energy intake. Moreover, we adjusted our models for energy intake to mitigate the effect of measurement error in the FFQs that individuals tend to misreport their intake of food and beverages in the same direction. Fourth, although we adjusted for several covariates including socioeconomic status, maternal psychopathological symptoms, lifestyle and child BMI, we cannot rule out genetic and unmeasured environmental confounding because of the observational nature of the study. Lastly, the non-response analyses suggested a possible selection bias, which may limit the generalizability of the study.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.