The attention-deficit/hyperactivity disorder-associated DRD4 7R allele predicts household economic status but not nutritional status in Northern Kenyan Rendille children

Abstract

Objectives

Around 11% of US children are diagnosed with attention-deficit/hyperactivity disorder (ADHD). One hypothesis for ADHD’s relatively high prevalence is that behaviors associated with ADHD were advantageous in past environments where they were positively selected for. A previous study showed that an allele associated with ADHD – the 7R allele of the gene encoding the D(4) dopamine receptor (DRD4) – had a positive effect on the nutritional status of nomadic adult Ariaal men and a negative effect on settled adult men. We attempted to replicate this finding by analyzing the impact of DRD4 7R on children’s nutrition and other household metrics in the Rendille, a population closely related to the Ariaal.

Methods

We genotyped 259 Rendille children aged 5 to 10 years old for DRD4 7R and analyzed this against previously collected anthropometric and household data from two Rendille towns. Analyses were pre-registered (https://osf.io/p8yv2/) before the addition of 7R genotype to the dataset.

Results

DRD4 7R was not associated with iron nutrition, indicated by transferrin receptor concentration, height-for-age or weight-for-height Z-scores, or with maternal education status. However, DRD4 7R was positively associated with household economic status (p = 0.047).

Conclusions

The failure to replicate an association between DRD4 7R and nutritional status might be due to this sample being of children who are not yet substantially provisioning themselves. Given that children’s genotypes are correlated with parents’ genotypes, it is likely that the effects of the parents’ genotypes, rather than the participating children’s, explains the association between children’s DRD4 7R genotype and household economic success.

Introduction

Attention-deficit/hyperactivity disorder (ADHD) is diagnosed when an individual shows a persistent pattern of inattention or hyperactivity-impulsivity that interferes with their functioning or development across multiple settings (American Psychiatric Association, 2013). ADHD is common, affecting an estimated 11.4% of US children between the ages of 3 and 17 (Danielson et al., 2024). When a ‘disorder’ is common, an evolutionary perspective suggests that the phenotype does not have a substantial detrimental impact on fitness, and/or that modern environments have changed and created a gene-environment mismatch (Stearns et al., 2010). ADHD is most prominently associated with poor academic achievement and inattention in a classroom environment (Nigg, 2013). Formal schooling is a distinctly modern phenomenon (Gray, 2011), and the behaviors associated with ADHD that may have once benefited fitness may negatively impact success in this and other modern structures. There is considerable speculation that ADHD is the result of this mismatch between the environments in which humans evolved and current novel environments (e.g. Hartmann, 2016; Jensen et al., 1997; Williams & Taylor, 2005). Evidence supporting this hypothesis has come from examining a polymorphism in the dopamine receptor D(4) (DRD4) gene.

A number of dopamine system genes have been found to be associated with ADHD (Gizer et al., 2009; Li et al., 2006). One locus of particular interest is the 48 bp variable number tandem repeat (VNTR) polymorphism located in exon III of the DRD4 gene. The 7-repeat (7R) allele of this polymorphism changes the functional and pharmacological properties of the D(4) dopamine receptor (Ferré et al., 2022; Misganaw, 2021). In vitro evidence shows that these changes can influence neuronal activation and inhibition, which may explain the behavioral differences between 7R and non-7R phenotypes (Homar-Ruano et al., 2023). The 7R allele has been shown to be associated with ADHD in multiple meta-analyses (Bonvicini et al., 2016; Bonvicini et al., 2020; Faraone et al., 2005; Gizer et al., 2009; Nikolaidis & Gray, 2009; Smith, 2010). Linkage disequilibrium analyses suggest that the 7R allele emerged approximately 50,000 years ago and subsequently underwent positive selection (Ding et al., 2002; Kidd et al., 2014; Wang et al., 2004; but see Naka et al., 2011). This timing of positive selection on DRD4 7R is before the emergence of agriculture and other components of modern environments and coincides with major expansions of modern humans out of Africa into different, novel environments that presented new challenges (McEvoy et al., 2011).

7R allelic frequencies are higher in populations that have migrated farther from Africa and are nomadic (Chen et al., 1999; Matthews & Butler, 2011; Tovo-Rodrigues et al., 2010). It has therefore been suggested that the 7R allele could confer fitness benefits to those expanding into new or unpredictable environments and living more nomadic lifestyles. Individuals with ADHD exhibit traits such as increased impulsivity and exploratory behavior that might be advantageous in variable environments that require quick responses or extensive searching for available resources (Jensen et al., 1997; Tovo-Rodrigues et al., 2010; Williams & Taylor, 2005). When a population is largely sedentary, the opposite may be true: traits like impulsivity and exploratory behavior may be selected against when individuals need to instead devote energy to acquiring resources from a familiar landscape (Chen et al., 1999; Williams & Taylor, 2005).

Further evidence of the 7R allele resulting in environmentally dependent fitness benefits has been demonstrated in the Ariaal, a traditionally nomadic pastoralist group in northern Kenya. In this study from Eisenberg and colleagues (2008), approximately half of the sample lived as nomads, whereas the other half had settled three decades prior. The 7R allele was associated with improved nutritional status in Ariaal men living nomadic lifestyles and worsened status in Ariaal men living in settled communities. This is consistent with the hypothesis that the 7R allele and its associated behaviors are more advantageous for nomadic individuals living in variable environments.

In this pre-registered study (https://osf.io/p8yv2/), we investigate the association between nutrition and DRD4 among 5–10 year old children in the Rendille, an ethnic group closely related to the Ariaal who reside on the same mountain slopes in the Marsabit district of northern Kenya (Shell-Duncan & Obiero, 2000). The names “Rendille” and “Ariaal” are fluid categories that frequently overlap. The Rendille, like the Ariaal, are traditionally nomadic. However, the communities studied here settled permanently in response to a series of droughts, and subsist primarily on sedentary cattle-keeping, agricultural goods, and market-centered commodities (Shell-Duncan & McDade, 2005). In both communities, children were facing high rates of undernutrition due to intense drought and food shortages at the time of data collection as well as high infectious disease loads (Shell-Duncan & McDade, 2005; Wander et al., 2009).

Based on the study conducted in the Ariaal, we expected the presence of at least one 7R allele to be associated with worsened nutritional status in children living in the settled communities studied here. This association could be caused by at least two distinct, but not mutually exclusive pathways. In the first pathway, a 7R allele could affect the behavior of the child directly. Behaviors such as impulsivity or hyperactivity could lead to less preferential treatment from caregivers or less success in self-provisioning. Hyperactivity and other ADHD-associated behaviors in children can also lead to poor maternal attention to that child, particularly during times of high stress (Befera & Barkley, 1985; Williams & Taylor, 2005). In the second pathway, a 7R allele may be associated with a child’s nutritional status because their genotypes proxy their parents’ genotypes. That is, if a child possesses a 7R allele, then at least one parent must also have a 7R allele. DRD4 genotype may be associated with the parents’ ability to obtain resources and provision their children. Thus, associations between DRD4 7R alleles and the nutritional status of a child could be due to the genotype’s direct influence of their child’s behavior or the behavior of their adult parents. To better distinguish between these two overlapping causal pathways, we examine how DRD4 relates to both household economic status and child’s nutritional status.

Methods

Study design and pre-registration

All phenotypic data were available and considered along with past studies of these data in designing these analyses. DRD4 genetic data are new and were not linked to phenotypic datasets while planning analyses. Hypotheses and planned analyses were agreed on by all co-authors and posted to Open Science Framework (https://osf.io/p8yv2/) before adding the primary predictor variable, DRD4 7R allele presence, into analyses. All analyses shown were pre-registered unless noted otherwise as post-hoc tests.

Study population and data collection

Data were collected in July 1999 with Rendille people in Marsabit District, Kenya (described in Shell-Duncan & McDade, 2005; Wander et al., 2009). The Rendille are traditionally nomadic, subsisting through camel pastoralism in the Kaisut Desert. This environment is very dry with less than 250 millimeters of rain per year (Nathan et al., 1996) and has high levels of endemic disease stress (McDade et al., 1998). Due to a series of droughts and associated loss of livestock, many Rendille have settled in permanent towns. Settlement was accompanied by dramatic shifts in lifestyle, including changes in diet and systems of subsistence (Shell-Duncan & Obiero, 2000). At the time of data collection, these communities were experiencing a particularly intense drought and were under an extreme level of nutritional stress.

The original study sample consisted of 318 settled Rendille schoolchildren from two communities, Korr and Karare. Korr is a lowland community located in the Kaisut Desert approximately 120 km west of Marsabit Town, the nearest administrative and economic center. Korr grew around a Catholic Mission distributing famine relief in the 1970s. At the time of data collection, Korr’s population numbered approximately 8,000, with several thousand more in neighboring areas. This region is too arid for agriculture, so residents lived by various means such as running shops, selling firewood collected from distant locations, brewing alcohol, or working for schools, development agencies, churches, or buying and selling livestock (Nathan et al., 1996). Karare is a sedentary highland community on Mt. Marsabit with a population of around 2,500 at the time of data collection. It is located along a graded road approximately 30 km from Marsabit Town. The main production systems in this community included sedentary cattle-keeping, involving animal and milk marketing in Marsabit Town, and dryland agricultural production of maize.

The data and blood samples utilized in this study were originally collected as part of a project which examined childhood nutrition and micronutrient deficiency in school-age children (5–10 years old). A full description of sampling is available in Shell-Duncan and McDade (2005) and Wander et al (2009). Researchers conducted a stratified randomized sampling strategy where they split each village into strata of equal population sizes of 5- to 10-year-old children and selected children randomly from household rosters. The study protocol was reviewed and approved by the Human Subjects Division at the University of Washington and the Ethics Committee at Kenyatta Hospital in Nairobi.

Nutritional status

Height was measured to the nearest millimeter with an anthropometer while the subject stood on a level platform. A Seca (Hanover, Md., USA) electric digital LED scale was used to measure weight to the nearest 0.1 kg, with the subject wearing light clothing (Shell-Duncan & McDade, 2005). Sex-specific height-for-age Z-scores (HAZ) and weight-for-height Z-scores (WHZ) were calculated in EpiInfo using CDC growth references from 2000 (EpiInfo version 1.0.5, Centers for Disease Control and Prevention, Atlanta, GA, USA). HAZ were also calculated against the more recent 2006 WHO standards for comparison using the ‘anthroplus’ package in R (de Onis et al., 2007). Iron status was assessed by measuring the concentration of transferrin receptor (TfR) in dried blood spot (DBS) samples using previously validated enzyme immunoassay protocols (McDade & Shell-Duncan, 2002).

Biological Sampling and Genotyping

Sterile, disposable lancets were used to collect free-flowing capillary blood (Shell-Duncan & McDade, 2004). At least two drops of whole blood were collected on filter paper (Schleicher & Schull #903, Keene, NH, USA). DNA was extracted from the dried blood spots (DBS) with the Gentra Puregene kit (Qiagen) using a modified protocol. Modifications included using three 1/8 inch punches (punched using a manual hole punch sanitized with 10% bleach between samples) rather than 50 μl of dried blood, centrifuging samples with Protein Precipitation Solution for 5 minutes rather than 3 minutes, and the substitution of DNA Hydration Solution with PCR grade water. All other steps were consistent with the Puregene protocol.

The DRD4 48 base pair repeat polymorphism was amplified by PCR using the protocol described in Eisenberg et al. (2008) with a few modifications to allow for smaller volumes of DNA per reaction. The PCR consisted of 1X Q-Solution (Qiagen), 1X buffer (Qiagen), 1 μM primer 1 (5′GCGACTACGTGGTCTACTCG3′), 1 μM primer 2 (5′AGGACCCTCATGGCCTTG 3′), 200 μM dATP, 200 μM dTTP, 200 μM dCTP, 100 μM dITP, 100 μM dGTP, 0.3 U of HotStar Taq (Qiagen), and 1 μl of DNA template, in a total reaction volume of 10 μl. The PCR profile began with 15 min at 95°C for enzyme activation and denaturing of template DNA followed by 40 cycles consisting of 1-min denaturation at 94°C, 1-min annealing at 55°C, and 1.5-min extension at 72°C; it finished with a 10-min extension at 72°C. Amplicons were electrophoresed through 2% agarose gels containing SYBR Safe (Invitrogen). Genotypes were determined after being compared to 50 bp ladders (New England Biolabs). Genotyping at the DRD4 locus can be vulnerable to allelic dropout which may result in the misclassification of heterozygotes as homozygotes (Eisenberg et al., 2007, 2008; Hamarman et al., 2005). Thus, the 120 apparent homozygotes were reanalyzed with fresh PCR reactions at 2X and 1/40X DNA template dilutions to confirm that samples scored as homozygotes were not heterozygous. This procedure did not reveal any evidence of allelic dropout.

Statistical Analysis

Hardy-Weinberg equilibriums for all individuals and for both communities were calculated in R using the ‘gap’ package (Zhao, 2007). For the primary analyses, the DRD4 genotype is coded as 7R allele present (noted as 7R+; includes 7R homozygotes and heterozygotes) or 7R allele absent (7R-), in line with previous studies (Eisenberg et al., 2008; Nikolaidis & Gray, 2009). For covariates, we include age, sex, community membership (Korr or Karare), economic status, maternal education, and the interaction of economic status and maternal education (economic status*maternal education).

Income in the form of cash or wages was uncommon in Korr and Karare at the time of data collection, so economic status of the household was determined through in-depth interviews and self-reports from mothers regarding ownership of assets (e.g. livestock), market and bartering activity, and wage income. This information was then dichotomized into “poor” and “economically sufficient” for each household. Since few mothers have more than a primary education, maternal education is categorized as ever attended school and never attended school. An interaction between economic status and maternal education was included because it was associated with WHZ in a previous study with Rendille children of which our current sample was a part (Shell-Duncan & Obiero, 2000).

Our outcome variables included height-for-age Z-score (HAZ), weight-for-height Z-score (WHZ), and transferrin receptor (TfR) concentration to investigate the effect of DRD4 on nutritional status. Height-for-age Z-score (HAZ) likely reflects prolonged or long-term nutritional status and is less likely to be impacted by transient or recent nutritional stresses. Weight-for-height Z-score (WHZ) is responsive to short-term changes in energy availability. Lastly, TfR is a biomarker of iron nutrition that is not influenced by infection (Wander et al., 2009). TfR expression, measured via soluble TfR, increases with tissue iron stress, therefore allowing us to infer a child’s access to nutrient dense, iron-rich foods – especially meats (Shell-Duncan & McDade, 2005). Investigating these together provides a fuller and locally salient picture of how DRD4 7R may influence nutrition among these children, particularly because they were experiencing an intense drought and food shortage at the time of data collection.

To test the hypothesis that DRD4 genotype is associated with nutritional status in these children and explore how potential associations are influenced by genetic effects on both the child and the parents, we ran a series of multiple linear regressions to test if 7R presence is associated with HAZ, WHZ, and TfR (Table 2). We performed these models while controlling for community membership, age, and sex (Table 2: Models 1, 3, and 5), and then controlling for economic status, maternal education, and the interaction of economic status and maternal education (economic status*maternal education) (Table 2: Models 2, 4, and 6). Conducting these regressions (first with individual-level covariates and then with individual-level and household traits) helps us to better determine how any observed associations between 7R presence and our measures of nutritional status occur through DRD4’s relationships with children’s cognitive/behavioral phenotype and their parents’.

Table 2.

Regression models predicting height-for-age Z-scores (HAZ), weight-for-height-Z-scores (WHZ), and transferrin receptor (TfR) concentration in Rendille children

Variable	HAZ		WHZ		TfR
	1	2	3	4	5	6
7R allele presence	0.147	0.174	−0.067	−0.039	−0.429	−0.536
Sex	0.103	0.111	−0.007	−0.002	−0.127	−0.162
Age	−0.014	0.001	−0.006	−0.005	−0.282*	−0.298*
Community	−0.109	−0.076	0.117	0.131	0.333	0.281
Economic Status	−	−0.313	−	−0.029	−	0.510
Maternal Education	−	0.950 **	−	−0.049	−	−0.221
Economic Status*Maternal Education	−	−0.788+	−	−0.221	−	0.961

Values are β coefficients;

⁺p < 0.10,

^*p < 0.05,

^**p < 0.01,

^***p < 0.001; All models are multiple linear regressions; N’s: HAZ - 258, WHZ - 259, TfR – 244

We also sought to understand the relationship between the DRD4 7R allele and household traits. Thus, we conducted two logistic regressions with household economic status and maternal education as the outcome variables in order to evaluate our hypothesis that DRD4 genotype predicts household traits. For maternal education, we included 7R presence, community membership, maternal age, and economic status as predictors (Table 3; ME). For economic status, we included 7R presence, community membership, maternal age, number of dependents in the household, and maternal education as predictors (Table 3: ES).

Table 3.

Logistic regression models predicting household variables in 245 individuals

Variable	Maternal Education	Economic Status
7R allele presence	1.233	1.760 *
Community	0.601	0.781
Maternal age	0.841 ***	1.020
Economic Status	1.705	−
Number of dependents in household	−	1.008
Maternal Education	−	1.526

Values are Odds Ratios;

⁺p < 0.10,

^*p < 0.05,

^**p < 0.01,

^***p < 0.001

We examined all regression models for multicollinearity by both noting the inter-correlations of all predictor variables and measuring variance inflation factors (VIFs), a measure of multicollinearity. Common cutoffs for collinearity concerns are ≥0.5 for pairwise correlations and ≥10 for VIFs (P. Vatcheva & Lee, 2016).

Results

Of the 318 individuals in the original study, ten were excluded due to substantial missing data and an additional 49 due to insufficient dried blood spot sample volume. This left a final sample for our analyses of 259 Rendille children from the two communities: 144 from Korr and 115 from Karare. The sample ranged from 5- to 10-years of age and had a mean age of 6.9 (SD = 1.53; Table 1). Means and standard deviations of height-for-age (HAZ) and weight-for-height (WHZ) Z-scores calculated using the 2000 CDC growth reference data are presented in Table 1. Total sample and sex-specific HAZ means were similar to HAZ calculated using current WHO growth reference data (see Supplementary Table S1). Further descriptive statistics can be found in Table 1. At least one copy of the 7R allele was found in 32.0% of the children sampled (Table 1). Hardy-Weinberg equilibrium was violated for both communities (Korr, p = 0.02; Karare, p = 0.03), and when both communities are combined (p = 0.002). Allele and genotype frequencies are reported in Supplementary Table S2. No multicollinearity was observed.

Table 1.

Descriptive statistics of variables of interest.

7R presence (% with at least one 7R allele)	32.0%
Males	31.3%
Females	32.8%
Sex (% male)	50.6%
Community (% Korr)	55.6%
Males	50.4%
Females	60.9%
Economic status (% economically sufficient)	44.8%
Males	41.2%
Females	48.4%
Maternal education (% with schooling)	17.8%
Males	16.8%
Females	18.8%
	Mean (SD)
Age (years)	6.9	(1.5)
Males	6.8	(1.6)
Females	6.9	(1.5)
Maternal age (years)	35.9	(10.3)
Males	34.3	(8.8)
Females	37.4	(11.4)
Number of dependents in household	3.4	(1.6)
Males	3.6	(1.7)
Females	3.3	(1.5)
Height-for-age Z-score (HAZ)	−1.4	(1.4)
Males	−1.5	(1.6)
Females	−1.4	(1.2)
Weight-for-height Z-score (WHZ)	−1.4	(0.8)
Males	−1.4	(0.8)
Females	−1.4	(0.8)
Transferrin receptor (TfR; mg/L)	5.3	(2.8)
Males	5.4	(3.1)
Females	5.2	(2.5)

N = 259. SD = Standard deviation.

Contrary to our predictions, 7R allele presence was not associated with HAZ, WHZ or TfR concentration (Table 2; all p > 0.16) or with maternal education (Table 3; p = 0.59). As a post-hoc test, we investigated the interaction between age and 7R in predicting nutritional status. Younger children’s nutritional status is likely dependent on parental provisioning, whereas older children may self-provision more, meaning the effects of the 7R allele may increase with age. Contrary to this prediction, we found no evidence of an interaction between age and the 7R allele (Supplementary Table S3).

We observed an association between the presence of a 7R allele and household economic status: households of children with at least one 7R allele were 1.8 times more likely to be economically sufficient than the households of children who did not have a 7R allele (Table 3; p = 0.047). In a post-hoc analysis we examined the dose-response association between number of 7R alleles and economic status and found, contrary to expectations, that having two 7R alleles did not increase the odds of living in an economically sufficient household over having one 7R allele (Figure 1; Supplementary Table S4).

Figure 1:

Bar chart displaying percentage of individuals in households with low versus high economic status grouped by number of 7R alleles.

Discussion

We expected that the 7R allele of the DRD4 gene would predict nutritional status in this sedentary environment. However, we did not find any associations between the 7R allele and the nutritional indices included in our analysis. This finding contradicts previous results showing that the 7R allele predicted worse health outcomes in adult males in a similar settled context (Eisenberg et al., 2008). This failure to replicate the previous findings might be explained by the children’s phenotypes having limited impact on their own nutritional status.

Our analyses found a positive association between the presence of the 7R allele and household economic status: children with a 7R allele were more likely to live in economically sufficient households. Post-hoc, this observation is somewhat in line with our predicted second pathway through which a 7R allele could influence nutritional status. In this case, the child’s genotype is a proxy for their parents’ genotype, whose behavior is facilitating greater economic success, although no positive impact on the nutritional status of the child is observed in our analysis. Given the relatively recent settling of the Rendille community at the time of data collection, this association could reflect economic sufficiency attained as nomads prior to settling, potentially due to behavioral advantages conferred by a 7R allele. This argument is complicated by the fact that the history of droughts in the area as well as the transition to a sedentary lifestyle can have varied effects on household economic status (Fratkin et al., 1999). Additionally, there are numerous complex behavioral and environmental factors that may contribute to household economic status in this region. While we suggest that DRD4 genotype may influence household economic success via behavioral phenotypes, the contribution of the 7R allele of DRD4 to this outcome is likely small and interacts in complex ways with other important factors such as drought or other seasonal conditions, the contributions of other members of the household, the patrilineal inheritance of assets, and the support of neighbors and kin (Fratkin et al., 2004; Fujita et al., 2004; Shell-Duncan & Obiero, 2000).

While geographically close to each other, the Ariaal and Rendille exhibit substantial differences in regard to their economic conditions and subsistence strategies. Further, the particular communities included in this analysis differ from the Ariaal communities included in Eisenberg et al. (2008) in substantial ways. The settled community included in Eisenberg et al. (2008), Songa, was largely dependent on irrigation agriculture, whereas most households relied on both cattle-keeping and subsistence agriculture in the two settled communities included here, Korr and Karare, at the time of data collection (Fratkin et al, 2004; Shell-Duncan & McDade, 2005). Individuals within Korr and Karare engage with market economies in varying degrees, as is common during a transition from pastoralist subsistence strategies towards sedentism. Although we did not propose this as a preregistered hypothesis, it is possible that lack of association observed between DRD4 7R and nutritional status is due to the allele being negative only in the context of an agricultural community such as Songa and neutral, or minimally beneficial, in communities which rely on cattle-keeping but which are not nomadic. Additionally, transitions to sedentism are economically complex and typically accompanied by changes in diet, which can have considerable impacts on nutrition and may overshadow any effect of the 7R allele, positive or negative (Fratkin et al., 2004; Shell-Duncan & Obiero, 2000).

This pre-registered study represents our attempt, using a dataset of schoolchildren from a closely related population, to replicate a previously found association between DRD4 7R and nutrition among adult Ariaal men. Our results do not support the hypothesis that DRD4 7R and its subsequent behavioral effects would decrease success in settled contexts. The failure to replicate this past result is potentially due to the current sample including only children who do not yet substantially influence their own nutritional status. However, the observed association between the 7R allele and economic sufficiency may be an indication that the 7R allele was beneficial prior to settling. In order to form any conclusions regarding potential fitness benefits conferred by the 7R allele, more study is needed to understand the effects of the allele in different contexts. The relationship between the behavioral effects of the 7R allele and their potential benefits to nomadic individuals is unclear in the Rendille at this time.