Neurodevelopmental effects of childhood malnutrition: A neuroimaging perspective

Approximately one in five children worldwide suffers from childhood malnutrition and its complications, including increased susceptibility to inflammation and infectious diseases. Due to improved early interventions, most of these children now survive early malnutrition, even in low-resource settings (LRS). However, many continue to exhibit neurodevelopmental deficits, including low IQ, poor school performance, and behavioral problems over their lifetimes. Most studies have relied on neuropsychological tests, school performance, and mental health and behavioral measures. Few studies, in contrast, have assessed brain structure and function, and to date, these have mainly relied on low-cost techniques, including electroencephalography (EEG) and evoked potentials (ERP). The use of more advanced methods of neuroimaging, including magnetic resonance imaging (MRI) and functional near-infrared spectroscopy (fNIRS), has been limited by cost factors and lack of availability of these technologies in developing countries, where malnutrition is nearly ubiquitous. This report summarizes the current state of knowledge and evidence gaps regarding childhood malnutrition and the study of its impact on neurodevelopment. It may help to inform the development of new strategies to improve the identification, classification, and treatment of neurodevelopmental disabilities in underserved populations at the highest risk for childhood malnutrition.


Definitions and Epidemiology of Childhood Malnutrition
While the global incidence of malnutrition has declined over time, rates remain alarmingly high. In early 2020, the World Health Organization (WHO) reported an estimated 181.9 million malnourished children in developing countries ( United Nations Children's Fund (UNICEF) et al., 2019 ), with five million yearly deaths in developing countries in malnourished children under five years of age. Childhood malnutrition has both tremendous health and economic impact and leads to immense national cost expenditure. For example, in the UK, in 2007, public expenditure on disease-related under-nutrition was estimated at more than £13 billion per year ( Ray et al., 2014 ).
The root causes of malnutrition are multi-dimensional and include poverty, poor sanitation, crowding, infectious diseases, maternal depression, and child abuse and neglect ( Bhutta et al., 2017 ;Mangala and Subramanyam, 2014 ). Substandard hygiene and living conditions can lead to diarrhea, inflammation, and infectious diseases that may exacerbate or lead to malnutrition in the developing infant ( Bolick and Guerrant, 2019 ;Guerrant et al., 1992 ). The relationship between maternal mental health conditions, especially depression and anxiety, and poor infant growth and undernutrition in early childhood are also known and have been reported in Asia ( Patel et al., 2003 ;Rahman et al., 2004 ) and Barbados ( Galler et al., 2004( Galler et al., , 1998. However, a relationship between maternal mental health disorders and infant undernutrition was not observed in a study of stunted and underweight Ethiopian infants, suggesting that regional differences in patterns of infant rearing may play a role ( Medhin et al., 2010 ). Finally, the early cessation of breastfeeding, multiple births, and lack of national health insurance coverage in mothers was also associated with an increased risk of childhood malnutrition ( Aheto et al., 2015 ). Because of the frequency with which these conditions can co-exist with or lead to infant malnutrition, it is important to measure and adjust for these variables in research studies examining the long-term consequences of childhood malnutrition on cognition and brain development ( Galler, 1987 ).
The two most common forms of childhood malnutrition in the developing world are 1) wasting, a consequence of severe (SAM) or moderate (MAM) acute malnutrition, and 2) stunting, which can be seen in more chronic forms of childhood malnutrition. Each phenotype has a unique impact on the brain and behavioral development. Although only about half of all children under five live in low-and middle-income countries (LMIC), 75% of all wasted children and two-thirds of stunted children live in these regions.

Wasting
Wasting is the result of moderate-severe acute malnutrition associated with substantial weight loss or failure to gain weight. The term refers specifically to a child who is too thin for their height. This condition results from a severe lack of macronutrient and micronutrient intake, complicated by co-morbid inflammation, diarrhea, and infectious diseases that often accompany childhood malnutrition. Children suffering from wasting also have weakened immunity, and they face an increased risk of death, particularly in severe cases. These children require urgent refeeding, treatment, and care to survive. The long-term effects of wasting are significant, and survivors frequently face life-long detrimental effects on brain, behavior, and health outcomes.
There are two distinct clinical conditions associated with wasting (or SAM) ( Galler, 1984 ): 1) Kwashiorkor, or nutritional edema, occurs most often between ages 1 and 3 years in children whose diet is grossly deficient in protein. This condition is clinically defined as growth failure, muscle wasting, and bilateral pitting edema of the extremities. Hair is easily plucked, and hair and skin may be depigmented. 2) Marasmus, or severe wasting due to the insufficient intake of protein and calories, typically occurs in children under the age of 1 year and may accompany premature weaning. These children are irritable and have a loss of subcutaneous fat, muscle wasting, and wizened facial features. However, they do not have edema.
Although many longitudinal and clinical case studies continue to refer to these well-recognized conditions, the more recent WHO designations of SAM and MAM have now replaced the earlier marasmus and kwashiorkor terminology. In children six to 59 months, SAM is defined as having a mid-upper arm circumference of < 115 mm or weight-for-height < 3 Z scores or bilateral pitting edema, or both ( Williams and Berkley, 2018 ). Moderate acute malnutrition (MAM) is defined as weight for height Z-scores between -2 and -3 and a mid-upper arm circumference ≥ 115 mm and < length < 125 mm. The most susceptible ages for SAM are six to eighteen months when brain and growth, in general, are very rapid ( Williams and Berkley, 2018 ); however, it may also occur in infants under six months of age ( Kerac et al., 2015 ).
As of March 2019, 47 million children under five years of age suffered from wasting, and nearly 17 million were severely wasted ( United Nations Children's Fund (UNICEF) et al., 2019 ). The highest global prevalence of wasting was reported in Asia, with more than half of all wasted children globally living in South Asia, with 14.3% of children under five categorized as suffering from wasting. However, wasting and severe wasting are acute conditions that can change frequently and rapidly over a calendar year. This makes it difficult to determine reliable trends over time with the available data, and current reported rates may underestimate the numbers of impacted children.

Stunting
The most common indicator of chronic childhood malnutrition is stunting, which refers to a child who is too short for their age. The devastating effects of stunting can last a lifetime ( United Nations Children's Fund (UNICEF) et al., 2019 ). Stunting results from inadequate nutritional intake, both in utero and early childhood, leading to potential and stunted children may never attain their full height potential even after intervention. Even more concerning, their brains may never develop to their full cognitive potential, and these individuals often face learning difficulties in school, with subsequent decreased earning capacity ( Black et al., 2017 ;Grantham-McGregor et al., 2007 ) According to a 2020 Global Nutrition Report on malnutrition, nearly one in every four children under five is stunted ( Development Initiatives, 2020 ). This translates to approximately 149 million children under five years of age who suffer from stunting, a reduction of almost 17 million cases from the figures in 2012. At present, almost 55% of all stunted children live in Asia, which translates to more than 81 million children, and 39% of stunted children are in Africa. The latter is the only region that reported an increase in stunting, based on figures between 2000 and 2019. Despite the global reduction in stunted children in regions other than Africa, the rate of decline has been considerably slower than projected, which has prompted a need for more effective ways to combat stunting, especially in lower-income countries ( United Nations Children's Fund (UNICEF) et al., 2019 ).

Cognitive and Mental Health Outcomes of Childhood Malnutrition
The next two sections will focus on the long-term effects of proteinenergy malnutrition (SAM) as it impacts the developing child over the lifespan and will focus on studies with control groups. It has long been recognized that poor childhood nutrition and associated social and environmental factors, including poverty, crowding, maternal depression, low maternal IQ, and child maltreatment, strongly affect cognitive, language, and socio-emotional development ( Anoop et al., 2004 ;Walker et al., 2011 ). Most of the longitudinal studies that we cite have taken at least socioeconomic status into account, while others have incorporated a broader definition of the child's microenvironment.
The immediate effects of SAM on neurodevelopment, intelligence, and cognitive performance have been extensively studied in children in a range of at-risk populations in different geographical settings ( Galler, 1984 ). These children are seriously ill and recovering from the metabolic consequences of nutritional disorder. A critical period from the second trimester of pregnancy to two years of age, or the growth spurt period, is thought to represent the most vulnerable period of brain development ( Dobbing, 1982( Dobbing, , 1981. Malnutrition insults during this phase result in irreversible changes in cognition and behavior, even after the acute period of illness, and may impact the ability of the child to interact with its environment, i.e., "functional isolation ". While protein deficiencies are known to be especially impactful concerning the developing brain, micronutrient deficiencies in early childhood, including low iron levels, often co-occur with protein-energy malnutrition ( Pollitt, 1995 ) and may lead independently to poor cognitive and behavioral outcomes ( Algarín et al., 2013 ;Lukowski et al., 2010 ;Suchdev et al., 2017 ).
The long-term effects of early childhood malnutrition (SAM) have been reviewed by us ( Galler, 1984 ;Galler et al., 1996 ) and others ( Grantham -McGregor et al., 2014 ;Laus et al., 2011 ;Prado and Dewey, 2014 ). Most evidence for the permanent effects of severe postnatal malnutrition comes from a small number of prospective studies that have followed previously malnourished children and control children from early childhood into adolescence and young adulthood ( Hertzig et al., 1972 ;Hoorweg and Stanfield, 1976 ;Raine et al., 2010 ;Stanfield, 1993 ;Stoch and Smythe, 1976 ). Common findings across these studies, that include populations from Jamaica, Uganda, South Africa, and Mauritius, show reduced intellectual capabilities and IQ, poor academic performance, attentional deficits, and poor executive control. Recovery from early malnutrition has also been associated with schizotypal personality in young adulthood ( Venables and Raine, 2012 ) mediated by IQ at age 11, and increased Neuroticism and decreased Extraversion, Openness, Agreeableness, and Conscientiousness in middle adulthood, also closely linked to cognitive profiles of the previously malnourished subjects ( Galler et al., 2013 ). The long-term effects of malnutrition are associated not with the type and severity of early child malnutrition, and also with its duration. It is noteworthy that similar findings of cognitive and attentional deficits were reported in chronic but mildly malnourished Kenyan children ( Sigman et al., 1989 ) and also in stunted Jamaican children with chronic undernutrition, although an early childhood stimulation program reversed many of these cognitive and behavioral effects ( Grantham -McGregor et al., 2014 ).
The Barbados Nutritional Study (BNS) is a lifespan longitudinal study with a controlled design and unique characteristics that has followed a cohort with histories of moderate-severe malnutrition limited to the first year of life. During the 1960s, childhood malnutrition was one of the most prevalent public health problems in Barbados, and many children were hospitalized with marasmus or kwashiorkor, both forms of SAM. To address this problem, the newly formed Barbados government implemented mandatory reporting of childhood malnutrition cases and a comprehensive government-supported intervention program that provided subsidized foods, nutrition education, home visits, health monitoring, and medical care, and a preschool nursery program for the children and their siblings until 12 years of age ( Ramsey, 1979 ). The BNS was designed as a case-control study, comparing children with suffered from marasmus or kwashiorkor limited to the first year of life with healthy classmate controls, matched by age, gender, and handedness. Malnourished and control children had normal birthweights ( > 2500 g), good Apgar scores, and no histories of encephalopathic events during childhood. Because they participated in the intervention program, none of the index children had evidence of continuing malnutrition after the first year of life, and they achieved full catch-up in physical growth by adolescence ( Galler et al., 1987a ) although cognitive and behavioral outcomes continue to be impacted.
As children and adolescents, the previously malnourished children had impaired cognitive performance, lower IQ, and attention deficits ( Galler et al., 1987b( Galler et al., , 1983b( Galler et al., , 1983a. They also had poor performance on a national high school examination, largely accounted for by their impaired IQ and attention deficits throughout their early school years ( Galler et al., 1990 ). Cognitive deficits and behavioral problems, especially inattention, continued into middle adulthood ( Galler et al., 2012a ;Waber et al., 2014 ). Adult IQ was lower in the previously malnourished group and closely correlated with childhood IQ; attentional deficits, especially inattention, were present to age 45 years and more prevalent than impulsivity across the lifespan. Importantly, the educational and occupational achievements of the cohort were reduced, as well as their incomes, underscoring the long-term economic burden of the history of early childhood malnutrition in this population ( Galler et al., 2012b ). As noted above, environmental influences remain a major consideration in studies of childhood malnutrition and may contribute substantially to the compromised outcomes of children with histories of childhood malnutrition. Indeed, the microenvironment and social circumstances of the previously malnourished children were compromised relative to the healthy control group with fewer household conveniences. While these conditions themselves were independently associated with cognition and behavioral outcomes, they did not, however, mediate the association between early malnutrition and cognitive or behavioral deficits seen in this cohort ( Galler & Ramsey, 1985 ;Waber et al., 2011 ). The BNS has also examined the role of maternal depression and anxiety ( Salt et al., 1988 ) and child maltreatment as potential mediators of the childhood malnutrition effects ( Hock et al., 2018( Hock et al., , 2017. More recently, the BNS has reported that the effects of early malnutrition on cognitive performance and attention can extend to the next generation. Parental history of moderate to severe infantile malnutrition was associated with increased cognitive and attentional problems in their offspring who had never been malnourished themselves ( Waber et al., 2018 ). This transgenerational effect may be mediated by epigenetic changes documented in this cohort ( Galler & Rabinowitz, 2016 ;Peter et al., 2016 ).

EEG
The neural effects of childhood malnutrition have not been studied extensively and, to date, there are only a handful of statistically controlled studies that have been conducted ( Gladstone et al., 2014 ). Moreover, only a small number of studies have followed children beyond the period of acute malnutrition. Most work to date has been done using EEG. EEG studies on SAM survivors indicate irreversible neurological damage when the exposure occurred before the age of two years. Early research conducted in South Africa by Smythe (1976 , 1967 ), as well as Baraitser and Evans (1969) and Evans et al. (1971) , reported EEG changes in 20 children with histories of marasmus between 10-24 months of age and 20 well-nourished control children up to 15 years after their malnutrition exposure. These results are challenging to interpret as they are primarily case-studies and are based on relatively small sample sizes. In their study of South African children, Bartel et al. (1979) determined that there were reduced average frequencies, lower-alpha activity, and more slow-wave activity in 6 to 12-year-old black children who had experienced kwashiorkor in infancy (up to 27 months) when compared to their siblings, healthy neighborhood children, and white children of high socioeconomic status. This study included 120 children, and the inclusion of three control groups allowed the role of social and environmental factors to be assessed.
Recent work by our group, using qEEG, has confirmed significant differences in brain structure and function between a Barbadian cohort of 5 to 11-year-old SAM survivors who were malnourished in the first year of life and classroom controls, matched by gender, age, and handedness Taboada-Crispi et al., 2018 ). Relative to controls, the SAM survivors showed decreased alpha activity, suggestive of neurodevelopmental delay. . Taboada-Crispi and colleagues, using a topographic analysis, demonstrated significant differences in z spectra between the same previously malnourished children and controls. These differences included increased theta (3.51, 4.68, and 5.07 Hz), alpha 2 (13.28 Hz), and beta (13.67-18.36 Hz), and a decrease in alpha 1 (8.98 Hz) . These findings at a topographic level were congruent with those obtained through source (tomography) analysis apart from a decrement of beta (16.40 Hz) ( Bringas-Vega et al 2019 ). Similarly, Valdés-Sosa et al. (2018) sought to validate the biomarkers of cognitive impairment in previously malnourished children. Together, these qEEG studies cement the role of neuroimaging in diagnosing the neural correlates of cognitive dysfunction found in children exposed to early childhood malnutrition .
Changes in cognition and brain function of the BNS cohort have been documented across the lifespan. Altered ERP activity was found to be associated with impaired executive functioning in the previously malnourished Barbados cohort as late as ages 45-51 years ( Roger et al., 2019 ). In this study, the SAM survivors showed lower N2 amplitudes and corresponding omission error rates on a Go-No-Go task versus controls. As shown, altered brain development is evident in SAM children and may have long-lasting consequences. A recent pilot study at ages 45-51 confirmed accelerated cognitive decline in the SAM survivors, using standardized MoCA and MMSE measures ( Razzaq et al., 2020 ). qEEG findings at 5-11 years predicted those individuals who were at greatest risk for cognitive impairment 40 years later.
The usefulness of EEG monitoring in high-risk child populations is shown in a recent longitudinal study by Xie et al. (2019) that linked faltering growth (though not frank malnutrition) with impending damage to functional connectivity and the subsequent cognitive ability of stunted children from Bangladesh's urban population. The EEG profile in the low-beta and theta frequencies sampled among cohorts of 6-month and 36-month children were recorded and analyzed and repeated at 27 and 48-months. The findings suggest that at 6-months of age, stunted children show an electrical profile that would predict cognitive impairment at 27 months of age.

fNIRS
Functional near-infrared spectroscopy (fNIRS) is another neuroimaging tool to study the functional aspects of the brain and is especially useful in the assessment of very young children Gallagher et al., 2012 ;Lloyd-Fox et al., 2010 ;Pinti et al., 2018 ). Recent advances in fNIRS methodology have enabled its use in a series of LRS studies on the impact of poor nutrition and early adversity on brain development. While these studies primarily focus on young children, to date, there is a paucity of reports examining the long-term effects of childhood malnutrition using fNIRS. In a pilot study conducted in Guinea-Bissau, the cerebral blood flow of children suffering from MAM was measured as a biological indication of impairment in cognitive functions ( Roberts et al., 2017 ). The study showed that one-to three-year-old children treated with daily nutritional supplements over 11 weeks improved in terms of working memory. However, there was no improvement in the cognitive performance of five-to seven-year-old supplemented children. Cerebral blood flow was significantly correlated with performance on a sorting task in a combined group of intervention and control children.
Findings from fNIRS studies lend further support to the pivotal role of an enriched environment, as poverty or the social environment of the infants dictated the brain responses to social stimuli as early as six months old ( Perdue et al., 2019 ). These studies indicate the potential usefulness of fNIRS in field studies of childhood malnutrition and their possible role contribution in documenting the impact of interventions on cognitive function.

Nuclear Magnetic Resonance (NMR)
NMR imaging, such as MRI and MRS, has also been used to study the effects of childhood malnutrition on the brain. However, there are very few studies in LRS due to the high cost of equipment and limited availability of these technologies in areas where malnutrition is prevalent. Many of the published reports (as in the case of EEG studies) are predominantly case studies that lack adequate control groups and have, for the most part, only studied the progress of concurrently malnourished children during their hospitalization and not after recovery. One of the few controlled studies using MRI to document the neural effects of early childhood malnutrition in Chilean high school students showed evidence of marked cerebral atrophy, particularly decreased brain volume, in both males and females that were present up to 18 years after the malnutrition exposure ( Ivanovic et al., 2000 ).
However, a more recent study of 9-year-old Malawian SAM survivors, who were malnourished between 15-32 months of age, demonstrated that, while previously malnourished children had impaired cognitive function, especially in visual memory and visual attention, seven years post-discharge from treatment, there were minimal alterations in brain morphology based on MRI scans ( Lelijveld et al., 2019 ). Only a subset of the malnourished group (but no controls) underwent MRI scans, and 51% (25/49) showed evidence of an abnormality, the most common of which was sinusitis (43%; 21/49). Other detected abnormalities included gliosis (8%; 4/49) and chronic stroke (2%; 1/49). However, the odds ratio for SAM survivors of having any brain abnormality was not statistically significant (OR = 1.23 (95% CI: 0.62, 2.44) ( p = 0.55) ( Lelijveld et al., 2019 ).
In one of only a few MRS studies, Cakir and colleagues compared malnourished and control Turkish infants (2-36 months), showing that the Choline to Creatinine (Cho/Cr) ratio was higher in all examined brain regions (thalamus, basal ganglia, and white matter) in infants with malnutrition vs. controls ( Cakir et al., 2019 ). However, only the thalamic Cho/Cr ratio showed a significant difference statistically. No data were available after infancy or after recovery from the malnutrition episode and there are no MRS studies in older SAM survivors.

Ongoing Neuroimaging Projects on Malnutrition
There are currently several projects underway examining the effects of poverty and malnutrition on the brain, including the Brain Imaging for Global Health (BRIGHT) project, and the Bangladesh Early Adversity Neuroimaging Project, and, as described earlier, the Barbados Nutrition Study. One of these projects -the BRIGHT project -is a timely and apt initiative to explore the relationship between early environment and neurocognitive development, comparing the development of infants from high (UK) resource with those from the Gambia, an LRS Lloyd-Fox et al., 2019 ). This study is longitudinal and has now followed infants from the prenatal period to two years of age, making use of fNIRS and EEG measures. Similar efforts by research teams in Bangladesh ( Perdue et al., 2019 ;Turesky et al., 2019 ) and India ( Wijeakumar et al., 2019 ) are also in progress. In these projects, fNIRS, EEG, and MRI are being used in infants from LRS to study the impact of poverty and associated environmental risk factors on cerebral and behavioral development. In one recent study using MRI, the investigators found a greater intrinsic functional connectivity between the amygdala and the precuneus in impoverished Bangladeshi infants as compared to more affluent infants ( Turesky et al., 2019 ). However, it is important to note that stunting was not associated with this amygdala-precuneus connectivity difference.
The question of how malnutrition, which is often accompanied by poverty, could mediate or dampen the development of other neurocognitive domains remains unknown still ( Perdue et al., 2019 ). One study conducted in Indian infants and toddlers reported similar results to those of Turesky and colleagues, reporting altered brain networks due to early environmental adversity ( Wijeakumar et al., 2019 ). Again, the impact of malnutrition -rather than adversity -on neurodevelopment was not directly investigated. The authors of these reports could only infer from their studies in LRS that malnutrition could be a mediating factor in these developmental differences. Thus, given the prevalence of childhood malnutrition in developing countries, further research is neces-

Table 1
Gaps in knowledge relating to early childhood malnutrition and neurodevelopmental outcomes.

Problem or Question Studies Needed
Timing of the insult and its long-term effects: How do pre-conceptional, gestational and postnatal malnutrition impact brain development and what are the brain manifestations over the lifespan?
Longitudinal studies are urgently needed, examining the long-term neural effects of malnutrition at different stages of early development.
Nutrition requirements: What are the macro (protein) -and micronutrient requirements in early childhood that promote normal brain development?
Studies of nutrition requirements to support brain development over the lifespan are needed. Ecology and home environment: Children exposed to malnutrition are also exposed to other co-morbid conditions that may contribute to their outcomes. Are malnutrition-related biomarkers modified by the individual, family and societal context of the malnourished child?
Brain studies needs to assess and identify role of poverty, crowding, maternal depression/parenting and maltreatment on brain structure and function.
Gender differences: Need better characterization of gender effects on linkages between nutrition and brain function/development. Outcomes and interventions need to consider possible gender-specific effects of early insults. All brain studies need to include both genders. Key brain and behavioral outcomes: Can specific changes in brain function/structure be used to identify early biomarkers of childhood malnutrition? What are the specific links between these biomarkers of brain structure/ function and cognitive/behavioral outcomes over the lifespan?
Methodologies appropriate for field work in LRS, such as qEEG and fNIRS, are urgently needed to study brain function and to improve our understanding of nutrition-related cognitive, executive control and behavior problems. * * Early malnutrition and aging: Do cognitive and brain effects of early malnutrition contribute to early cognitive decline and dementia?
Brain studies may point to early mental decline and may allow for intervention at earlier stages of development. Risk and resilience: A better understanding of the variability in brain responses to early stressors can assist the development of targeted interventions in highest risk groups.
Longitudinal studies may identify subgroups of individuals who are more resilient to the adverse neurodevelopmental effects of early malnutrition.
Accessing other databases: Data from "natural experiments " arising from famine, prolonged and extreme dietary restriction and epidemics may already be available. These data can potentially provide important information in planning new clinical studies of early malnutrition and brain structure and function.
Identifying and accessing databases of existing longitudinal cohorts should be encouraged. Leveraging data from existing data sets will be an important adjunct to new longitudinal and collaborative studies in different geographic settings.
Underlying Mechanisms: ■ Epigenetic changes as a biomarker of impaired brain and behavior function and resilience. ■ Translational research allows direct examination of neural mechanisms in controlled laboratory settings and is complementary to clinical and epidemiological studies of early childhood malnutrition.
■ Epigenetic studies relating early nutritional deficits with epigenetic changes and their role in lifespan outcomes and intergenerational transmission of malnutrition effects. ■ Translational studies in animal models are urgently needed to examine mechanisms and causality.
sary to identify the specific contribution of malnutrition to neurocognitive development, in the context of the many early adversities experienced by children in these settings. Fortunately, the research community has recognized the significance of nutrition and other associated biological and environmental risk factors and has been undertaking efforts to conduct studies in regions with resource limitations, particularly in LMIC.

Research Gaps in Studies of Early Childhood Malnutrition and The Brain
Gaps in knowledge related to childhood malnutrition and neurodevelopment are summarized in Table 1 . The cognitive, behavioral, and brain consequences of moderate-severe childhood malnutrition may seriously limit the educational and occupational opportunities of impacted individuals throughout their lifespan ( Fink et al., 2016 ;Galler et al., 2012a ). There is therefore an urgent need to support systematic, welldesigned studies capable of identifying such risks as the basis for developing earlier and improved intervention strategies.
Recent advancements in the field of fetal neuroimaging, particularly in NMR applications, have paved exciting opportunities in our endeavor to bridge the gap in understanding the underlying mechanism of malnutrition and its effects on children and their brain and cognitive development. In addition to conventional MRI, Counsell et al. (2019) outlined advanced quantitative NMR imaging tools in fetal neuroimaging, a few of which may be translated into childhood malnutrition studies. In particular, functional MRI (fMRI) ( Goksan et al., 2017 ), diffusionweighted imaging (DWI) ( Lockwood Estrin et al., 2016 ), arterial spin labeling (ASL) ( De Vis et al., 2015 ), and magnetic resonance angiography (MRA) and venography (MRV) ( Lee-Jayaram et al., 2020 ;Miller et al., 2012 ) would be useful in neuroimaging childhood malnutrition, as those have been applied in fetal and pediatric neuroimaging. For example, the study of brain structural impact of malnutrition may benefit from DWI by outlining the integrity of white matter tracts and comparing them between malnourished and typically developing children. Alterations and impairments in cognitive developments of malnourished children can be detected using fMRI, as applied in pediatric fMRI using task-based ( Buck et al., 2020 ) and resting-state paradigms ( Tuerk et al., 2020 ). Furthermore, brain physiological studies can take advantage of arterial spin labeling to study the perfusion of the brain, as well as MR angiography and venography to study disorders in the blood supply system in those children as well ( Lee-Jayaram et al., 2020 ;Miller et al., 2012 ).
While many studies have examined the relationship between early childhood malnutrition and cognitive outcomes, in addition to the availability of advanced screening and imaging technologies ( Deoni, 2018 ;Jasi ń ska and Guei, 2018 ), the direct investigation of brain structure and function has been limited by the high costs and limitations of applying the latest technologies in LRS. As these strategies become more accessible in third-world countries and LRS, studies linking brain consequences to childhood malnutrition and associated adversities will be more feasible and may help to better identify suitable biomarkers for measuring the impact of the early nutritional insult. Moreover, longitudinal studies comparing different neuroimaging techniques in high-risk children will also be beneficial. Such studies are needed to improve the efficacy and impact of interventions and should also be considered a high priority.
Finally, there are gaps in our understanding of resilience and protective factors in disadvantaged populations that may modify the adverse malnutrition impact. Studies regarding human adaptation to malnutrition and other associated early life adversities are lacking. This point was emphasized in recent reviews of global perspectives on risk and resilience in children and its important role in global child developmental studies ( Masten, 2014 ;Walker et al., 2011 ) and in identifying potential strategies to limit mental health disorders in high-risk LRS children ( Betancourt et al., 2013 ).

Developing Interventions
The problem of malnutrition among children, especially those in rural and impoverished areas, is endemic. The situation is particularly worrying because malnutrition has long-term and potentially irreversible effects on the affected children's cognitive ability, intelligence quotient, and behavior. Therefore, early interventions are crucial to minimizing the incidence of preventable cognitive and behavioral disorders associated with early malnutrition. For this to be effective, neuroimaging biomarkers of cognitive and behavioral function are essential, so that accurate and focused intervention can be developed for those at the highest risk.
Although malnutrition can manifest in multiple ways, the paths to prevention are virtually identical: adequate maternal nutrition before and during pregnancy and lactation; optimal breastfeeding in the first two years of life; nutritious, diverse, and safe foods in early childhood; a healthy environment, including access to basic health, water, hygiene, and sanitation services; and opportunities for safe physical activity. These key ingredients can deliver a world where children are free from all forms of malnutrition ( United Nations Children's Fund (UNICEF) et al., 2019 ).
Once a child has suffered from malnutrition, nutrition interventions alone are not sufficient to reverse the undernutrition's effects. Because of the brain's rapid growth from the 2 nd trimester of pregnancy to two years of age ( Dobbing, 1990 ;Galler et al., 1996 ), nutritional deficits during this early prenatal and postnatal period are known to result in irreversible changes in the brain. Nutritional deficits in later stages of childhood and adolescence may also impact the brain  and have led to the theory that there are additional "sensitive periods " of brain development ( Berens and Nelson, 2015 ;Nelson, 1999 ;Nelson and Gabard-Durnam, 2020 ). The loss of developmental potential resulting from childhood undernutrition is immense ( Grantham -McGregor et al., 2014 ). In 2017, it was estimated that as many as 250 million children un-der five years of age worldwide failed to reach their full developmental potential, in large part attributable to inadequate nutrition ( Black et al., 2017 ).
Documenting the cognitive, mental health, and brain consequences of early childhood malnutrition are central to the successful development of targeted and cost-effective interventions. Moreover, integrated programs that include nutritional interventions, as well as social and cognitive stimulation, are more likely to provide maximal benefit ( Baqui et al., 2008 ;Black et al., 2008 ;Grantham -McGregor et al., 2014 ;Waber et al., 1981 ;Walker et al., 2007 ). However, even such integrated care, while effective for stunting, may not fully reverse the effects of malnutrition during critical periods of brain development ( Galler et al., 2013 ).
Finally, the strategies of interventions should be sensitive to local culture and cost-effective, send simple messages, and overcome widespread unawareness and illiteracy in developing countries. The strategies should include various stakeholders, namely, government, physicians, nutritionists, schools, and relevant non-governmental organizations. Furthermore, the integration of services by different governmental agencies, including ministries of health, nutrition, legal, agriculture, and education, is important and central to successful interventions.

Future Directions: Pediatric Neuroimaging Techniques
Malnutrition is a disorder that primarily affects children in developing countries ( Ngo et al., 2015 ). As we have discussed earlier, early screening, detection, and intervention are needed to circumvent the neurological impact and faltered growth of affected children. However, screening modalities remain severely limited in these countries. The main issues hindering the widespread screening are 1) lack of availability and accessibility, 2) lack of expertise, and 3) affordability ( Table 2 outlines the advantages, disadvantages, accessibility, and usage of neuroimaging modalities in developing countries and LRS. Advanced screening modalities are often only available in cities with tertiary healthcare facilities ( Ogbole et al., 2018 ), while malnutrition often occurs in rural and impoverished areas ( Ramokolo et al., 2018 ).
Looking at the rate of the past ten years, the utilization of neuroimaging in developing countries has indeed expanded, yet it remains restricted by cost factors. It is presumed that even with the rapid advances and globalization in research technologies, the use of pediatric neuroimaging modalities in developing or near-developing countries is mostly limited to EEG or, more recently, fNIRS, considering their portability and much lower cost . Nevertheless, in the upcoming 5-10 years, we suggest that EEG is perhaps the best choice for widespread, cost-effective initial screening in rural healthcare facilities. Additionally, improved qEEG methods are useful in identifying early neural biomarkers of malnutrition that can serve as the basis of developing targeted interventions in LRS modalities and can be easily measured across the lifespan. EEG is a relatively cheaper technology compared to others. For example, an fNIRS machine with 16 channels can cost around 72,000 USD, while a 16-channel EEG machine would only cost around 19,000 USD. The lower installation and operating costs would translate into lower screening costs, making the technology more accessible to the masses . For a rough overview, citing an example of epilepsy screening in Malaysia, 54% of children diagnosed with epilepsy received EEG screening, compared to only 12% who undergo MRI. EEG's diagnostic cost in the mentioned sample was reported to be around 2800 MYR, while MRI was in the 5000 MYR range ( Salih et al., 2012 ).
In summary, for the next 5-10 years, the most basic and costeffective neuroimaging technology that can be provided is EEG, but with some trade-offs such as low spatial resolution and vulnerability to an array of artifacts. However, it remains an excellent, objective screening tool to better identify and classify children suffering from early childhood malnutrition.

Conclusions
We have summarized key scientific studies and challenges arising from studies in developing countries on childhood malnutrition and its effects on the brain and cognition. These studies have been consistent in demonstrating a decline in cognitive ability over the lifespan. There are also lasting cognitive effects of childhood malnutrition that impact the next generation. However, neuroimaging studies using newer methodologies to examine brain structure and function in populations with histories of childhood malnutrition are limited, and findings to date have not been linked to cognitive and behavioral outcomes. A comprehensive picture of the effects of childhood malnutrition on the brain, linking brain and cognitive changes over the life span, is urgently needed. Such knowledge will improve our ability to identify individuals at the highest risk for neurodevelopmental disorders, thus providing opportunities to develop early targeted interventions to improve the health outcomes and quality of life for individuals who have suffered from childhood malnutrition. Importantly, this knowledge could provide evidence-based information that will inform public health policy in LRS.
Children are the most crucial resource for the next generation. Neuroimaging screening tools for large-scale assessments of brain function in the pediatric population at risk for malnutrition can serve to identify new biomarkers that can be linked to neuropsychological and behavioral findings. Screening should focus on children from birth to two years of age, during the most rapid period of brain development, with frequent follow-up assessments at least through adolescence and into early adulthood. EEG is a non-invasive and portable modality; it is the most cost-effective neuroimaging screening tool that has shown promising results in recording brain activity, and it may provide a better temporal resolution compared to other methods. The integration of EEG and fNIRS is also a promising approach, and these screening tools can provide better spatio-temporal resolution. This multi-modal integration is also non-invasive, portable, and cost-effective. The use of neuroimaging modalities to measure brain structure and functions as screening tools is feasible and cost-effective use in LRS and those populations at greatest risk for childhood malnutrition.