Skip to main content
PMC home page
Biological Research for Nursing logoLink to Biological Research for Nursing
. 2019 Aug 13;22(1):5–12. doi: 10.1177/1099800419869507

Genetic Risk Factors for Poor Cognitive Development in Children With Low Birth Weight

Lisa M Blair 1,✉, Rita H Pickler 2, P Cristian Gugiu 3, Jodi L Ford 2, Cindy L Munro 4, Cindy M Anderson 2
PMCID: PMC7068752  PMID: 31409118

Abstract

Low birth weight is an ongoing public health problem with severe consequences for those affected, including early morbidity and mortality and elevated risk for lifelong deficits in cognitive function. These deficits can be ameliorated by early intervention in many cases. To contribute to criteria for earlier identification of at-risk children prior to the onset of delays or deficits, we examined relationships between three gene candidates—SLC6A4, BDNF, COMT—and cognitive outcomes at school age in a secondary analysis of existing data from a nationally representative cohort. Single nucleotide polymorphism rs4074134, a variant of BDNF, and a rare insertion/deletion in the intron region of SLC6A4 were significant predictors of cognitive performance. Our final model predicted 17% of the variance in composite cognitive test scores among children with low birth weight at school age (F = 96.36, p < .001, R 2 = .17). Specifically, children homozygous for cytosine at rs4074134 scored .62 standard deviations higher on a measure of global cognition than children with one or more thymine. Similarly, children with an extra-long copy number variant of SLC6A4 scored .88 standard deviations higher than children who had one or more short forms of the gene. These findings support the potential for an approach to identifying children with low birth weights who are most at need of early intervention services. Future research should focus on validation of these findings in an independent sample and confirmation of the biological mechanisms through which these genes influence cognitive development.

Keywords: cognition, low birth weight, genetic, BDNF, SLC6A4

Despite decades of preventive efforts, low birth weight (LBW; weight < 2,500 g at birth) remains a major public health problem in the United States, occurring in more than 8% of all live births (Hamilton, Martin, & Osterman, 2018). There are major racial and regional disparities in the prevalence of LBW: African Americans experience LBW at a rate of over 13% of all live births (Hamilton et al., 2018), while some rural southern Appalachian counties report rates as high as 45% (Bailey & Cole, 2009). These disparities place major burdens on at-risk communities and populations and reduce the potential for affected individuals and families to overcome the adverse circumstances that are often associated with the incidence of LBW. The increased incidence of lifelong impairments to executive functions and domain-specific cognition is particularly concerning in survivors of LBW (Aarnoudse-Moens, Weisglas-Kuperus, Goudoever, & Oosterlaan, 2009). In a recent study of adults born at LBW (mean birth weight = 1,306.7 g), Kroll et al. (2017) demonstrated moderate-to-large effect size for impairments in executive function that were further correlated with poor outcomes in educational attainment, employment, and social function.

Researchers of early intervention programs have long supported the notion that cognitive outcomes can be improved for children with LBW and other perinatal risk factors (e.g., preterm birth, birth injury; Hilderman & Harris, 2014). One challenge to early intervention has been identifying children at greatest risk for cognitive impairment and enrolling them in early intervention at a time when there may be greater brain plasticity and, thus, greater opportunity to affect outcomes. Early intervention programs in the United States are resource limited and often rely on generalized criteria such as birth weight and gestational age to determine eligibility (Association of Maternal & Child Health Programs, 2013). Yet in communities with severe disparities in LBW, available programs may not be able to accommodate all children with LBW. Thus, an approach to identifying children at highest risk for developing cognitive impairments may improve efficiency of the use of early intervention resources.

Numerous recent, high-quality studies have reported links between genetic variations and neurodevelopmental outcomes across the life span. Our previous synthesis of the literature revealed associations between 43 unique polymorphisms and neurodevelopmental outcomes across the life span (Blair, Pickler, & Anderson, 2016), including many disorders that disproportionally affect those born preterm or at LBW. These genetic variations may be important mediators in the relationship between LBW and poor cognitive outcomes, yet little research has been done to explore these associations in relation to LBW infants, who may be more vulnerable to gene–environment interactions due to their physiologic fragility.

To assess whether genotype may be able to assist in early identification of children with LBW who are at higher risk for cognitive impairment, we selected candidate genes and gene variants based on four criteria: (1) biological plausibility at the gene level to affect cognitive development, (2) availability in the selected preexisting data set, (3) identified as a likely candidate based on the results of our previous integrative review (Blair et al., 2016), and (4) identified as a likely candidate in an updated scoping search of the literature to examine findings specific to associations between cognitive effects and the proposed genes in infants. Based on these criteria, we identified three gene targets: SLC6A4, BDNF, and COMT. Specifically, the variants of these genes available in the preexisting data from the Fragile Families Study were a 44-bp insertion/deletion in an intronic region of SLC6A4 (also known as 5-HTTLPR), a single nucleotide polymorphism (SNP) at location rs4074134 of BDNF, and a SNP at location rs4680 of COMT. Literature on the effects of specific polymorphic variants was limited, so we evaluated biological plausibility at the full gene level where necessary.

To examine the role of these candidate genes in the development of cognition in children with LBW, we conducted an exploratory secondary analysis of existing nationally representative data with the goal of identifying potential targets for a precision health approach to cognitive risk screening and early intervention referral.

Candidate Genes

SLC6A4

The SLC6A4 gene (solute carrier family 6) is located on chromosome 17q11.2 and encodes a protein known as 5-hydroxytryptamine transporter (5-HTT), also known as serotonin transporter (OMIM, 2016b). Serotonin is an important neurotransmitter in both central and peripheral nervous system function. 5-HTT is responsible for transporting serotonin from the synapse into the presynaptic neuron; thus, 5-HTT terminates the activity of serotonin in the synapse and makes the serotonin available for reuse. Serotonin reuptake, or transport of serotonin back into the presynaptic neuron, plays an important role in psychopathologic conditions including major depressive disorder and anxiety disorders and is the target for selective serotonin reuptake inhibitor antidepressants. The polymorphism we examined in the present study is a 44-bp insertion/deletion copy number variant with 14 repeats on the short (S) allele and 16 repeats on the commonly occurring long (L) allele. The S form has been shown to produce about 60% as much transcription action as the L form and therefore is considered the low-performing allele. Frequencies of S-form have been reported to range from 13% to 44% and to be less common in populations of African descent (Lotrich, Pollock, & Ferrell, 2003). Additional alleles have also been described in populations of African descent that are not present in European ancestral lines (Vijayendran et al., 2012).

In mouse models, early-life serotonin transporter suppression has been associated with abnormal emotional function, indicating that low-performing SLC6A4 alleles (including the S-form) may play an important role in maturation of brain systems common to emotional function and learning (Ansorge, Zhou, Lira, Hen, & Gingrich, 2004). Esaki et al. (2005) reported that 5-HTTP-knockout mice exhibited lower glucose utilization in brain tissues and lower response to stimuli than wild-type mice. This finding suggests that brain glucose metabolism, and therefore growth and performance, may be altered by serotonin-transporter availability. In addition, interaction effects between SLC6A4 and the autism candidate gene Pten, where low-performing variants of SLC6A4 exacerbated the negative effects of Pten alone on phenotype (i.e., reduced sociability and macrocephaly), have been demonstrated in animal models (Page, Kuti, Prestia, & Sur, 2009). Cumulatively, these animal models demonstrate early-life effects of serotonin-transporter availability on brain development, though the phenotypes studied have been largely psychopathological and structural in nature.

BDNF

The BDNF gene is located at 11p14.1 and codes brain-derived neurotrophic factor (BDNF) protein. BDNF protein is required for the survival of striatal neurons and is one of several neurotrophic factors capable of binding to nerve growth factor receptors. Multiple phenotypes are associated with BDNF genetic variation and protein expression, including susceptibility to impaired memory and eating disorders (OMIM, 2017). The variant of interest in this study is a (G to A) SNP in an intron: rs4074134. Reported minor allele frequencies range from 0.11 to 0.56 across ancestral groups (OMIM, 2017). The location of rs4074134 in an intron may alter gene expression through gene splicing, a possibility supported by the empirical evidence of relationships between the variant and multiple behavioral phenotypes, though this has not been confirmed molecularly.

SNP rs4074134 has previously been associated with modifications in addiction behaviors related to tobacco use and with obesity (Tobacco & Genetics Consortium, 2010; Zhao et al., 2009). Research has empirically linked this SNP to susceptibility to obesity (Herrera, Keildson, & Lindgren, 2011) and alterations in smoking behaviors (Tobacco & Genetics Consortium, 2010) across multiple populations, but it has not been studied in relation to cognition. However, the action of BDNF on brain development is relatively well known. Extensive BDNF expression has been observed in fetal brain tissue and in adult brain structures including the hypothalamus, amygdala, fontal cortex, cerebellum, and pituitary gland (AceView, n.d.-b).

COMT

COMT, located on 22q11.21, encodes for catechol-o-methyltransferase (COMT), an enzyme responsible for the degradation of the neurotransmitters dopamine, epinephrine, and norepinephrine, and is important in the regulation of estrogen and pregnancy (AceView, n.d.-a). Phenotypically, the COMT gene has been characterized as clinically important in the use of catechol medications including dopaminergic drugs used to treat neurological conditions. Furthermore, COMT has been well studied as a susceptibility gene for schizophrenia and panic disorder. SNP rs4680 (G to A) is a missense change, meaning that the amino acid valine is substituted by methionine, influencing the enzyme activity of COMT. Global minor allele (A) frequency is 36%, but wide population variation has been noted. SNP rs4680 is a strong candidate gene for schizophrenia, Alzheimer’s disease, and Parkinson’s disease due to its role in dopamine regulation (OMIM, 2016a).

Expression of the COMT protein has been found in placental tissue and in various brain structures including the hypothalamus, corpus callosum, and medulla (AceView, n.d.-a). Sex-specific effects of COMT on emotional regulation and disease risk have been described in both animal and human populations (OMIM, 2016a). Specifically, in female but not male mice, COMT protein deficiency resulted in impaired emotional reactivity (Gogos et al., 1998). In humans, Alsobrook et al. (2002) found an association between a COMT polymorphism and obsessive–compulsive disorder in female but not male patients.

Method

We selected the sample for this secondary analysis of existing data from the Fragile Families and Child Wellbeing Study (Fragile Families), a longitudinal birth cohort study of children and their families (Center for Research on Child Wellbeing, 2015). Between 1998 and 2000, investigators recruited 4,898 families (each with a focal child, mother, and father when available) at the birth of the focal child from hospitals in 20 large, urban cities in the United States (populations > 250,000). The Fragile Families cohort contains a subset of families from 16 U.S. cities that, with the appropriate application of sample weights to correct for the complex sampling design, are nationally representative of children born in 1998–2000 in large U.S. cities. Data waves were conducted at the child’s birth and approximate ages 1, 3, 5, and 9 years, with collection of parent surveys, teacher surveys, caregiver/day-care surveys, biological measures, observations, geographical information, and genetic samples from mother and child. Overall sample attrition by the 9-year data collection was approximately 30%. For the purposes of the proposed analysis, we included children if they were in the nationally representative subsample, completed at least one cognitive measure at age 9, provided saliva for DNA analysis, and weighed <2,500 g at birth by maternal report (n = 169). We eliminated children of multiple gestation and one child with >30% missing data on variables in our multiple imputation model. The Ohio State University Biomedical Institutional Review Board (IRB) approved this study. Mothers and fathers provided written consent at the time of enrollment and children provided assent at age 9.

Independent Measures

Fragile Families investigators collected DNA from focal children using Oragene™ self-collection kits during the 9-year in-home observation. They processed self-collection kits with a median yield of 100 μg DNA over a 2-year period using the Oragene Laboratory Protocol Manual Purification of DNA. Genotyping methodology has been described in detail previously (Bendheim-Thoman Center for Research on Child Wellbeing & Department of Molecular Biology, 2016). Investigators at the Department of Molecular Biology at Princeton University performed the genotyping. Briefly, the SLC6A4 insertion/deletion was genotyped using polymerase chain reaction (PCR) followed by gel electrophoresis to identify copy number variants (S = short, L = long, X = rare variant longer than L). The SNPs rs4680 and rs4074134 were genotyped using TaqMan™ PCR and SNP Genotyping Assays from Applied Biosystems. Data were analyzed using the Applied Biosystems Sequence Detection Systems V.2.3 for 7900HT Reverse Transcriptase polymerase chain reactoin (RT-PCR) machine. Quality control was performed using Dnase-free H2O and three different positive controls (FAM, VIC, and both) with every plate for SNP genotyping. A lab technician and the primary genomic investigator reviewed positive (double-banded sample) and negative (Dnase-free H2O) controls for the SL6CA4 gels.

Genetic variants were coded nominally in the original data. That is, each unique genotype (i.e., 1 = “CC,” 2 = “CT,” 3 = “TT”) was assigned a code based on its frequency within the full sample. We converted these to dummy codes for analysis, such that each genotype was assigned a single-item, dichotomous condition to enable interpretation. For the copy number variant of SLC6A4, specifically, multiple rare allelic forms were noted within the sample, including traditionally measured “short” (44-bp deletion from ancestral allele) and “long” (ancestral allele) varieties as well as multiple variants shorter than “short” and longer than “long.” As the mechanism of action of these rare variants is not well characterized in the literature, we made the a priori decision to drop any genotype with <.05% frequency in the current sample to prevent outlier bias (n = 1). This resulted in four retained genotypes for SLC6A4: L/L, S/S, L/S, and X/L.

Dependent Measures

Children were assessed on four measures across multiple domains of cognitive performance and achievement. The Woodcock Johnson Passage Comprehension (subtest 9; WJ9) has raw scores ranging from 0 to 40 and tests reading and verbal (printed) language comprehension (r 1 = .88 on normative sample; Schrank, McGrew, & Woodcock, 2001). The Woodcock Johnson Applied Problems (subtest 10; WJ10) has raw scores ranging from 0 to 54 and tests quantitative reasoning, math achievement, and math knowledge (r = .93 on normative sample; Schrank et al.). The Wechsler Intelligence Scale for Children Digit Span subtest is considered a partial measure of general intelligence; it has raw scores ranging from 0 to 32 and tests short-term memory, sequencing, attention, and concentration using a 16-item scale (α = .85, corrected r = .83 on normative sample; Williams, Weiss, & Rolfhus, 2003). The Peabody Picture Vocabulary Test (PPVT) is a dichotomous measure of receptive vocabulary and verbal ability (α = .95 on normative sample), with raw scores ranging from 41 to 185 (PPVT, Fourth Edition, n.d.). Item-level measurements on cognitive tests are not available for further psychometric testing due to restrictions on the Fragile Families data. As normative samples were not representative of an LBW population, we constrained our analysis to raw test scores to prevent the introduction of bias from standardization on a nonrepresentative sample. Raw scores were treated as continuous variables.

To improve our understanding of the effect of genetic factors on overall child cognition, we created a composite score from the four raw test scores. We first assessed each measure for univariate normality and performed Jones-Pewsey transformations to improve normality on WJ9 and WJ10. Following transformation of those two variables, all four measures were appropriately normal on visual plot inspection and had skewness and kurtosis statistics <|1|. We conducted a parallel analysis to confirm unidimensionality and then generated standardized composite scores using an exploratory factor analysis. As genetic contributions to cognition have previously been noted to have sex-specific effects, we included biological sex (measured by maternal report at birth) in all full models.

Analysis

A power analysis completed with G*Power 3.10 indicated that our sample was sufficient to detect small to moderate effect sizes. Multiple imputation of missing data from the Fragile Families study, including sociodemographic characteristics (other than household income) and dependent measures, was completed using a method sensitive to levels of measurement. No variable imputed contained >20% within-wave missingness, and a single participant (n = 1) with >30% within-wave missing was dropped prior to imputation. Genetic factors were not imputed.

A multiple regression was performed using SAS Version 9.4.3 surveyreg procedure to adjust for the complex sampling design of the original survey, with all regressions including the maternal baseline national and replicate weights. Carriers of the minor BDNF “T” allele, the minor COMT “A” allele, and the “S” or short form of SLC6A4 were designated as the reference groups. We further report the coefficients of each of the four original measures of cognition as a secondary finding to allow for test-specific comparison across samples without unduly inflating the Type I error of our overall study.

Results

We have provided descriptive statistics for the analytic sample in Table 1. Briefly, 175 children with LBW in the nationally representative sample provided saliva for DNA analysis and had results of cognitive testing on at least one measure at age 9 years. ANOVA results are presented in Table 2. Our composite model was statistically significant, with an overall ability to predict 17% of the variance in composite test scores.

Table 1.

Descriptive Statistics for Analytic Sample of Children With Low Birth Weight Who Were Part of a Nationally Representative Sample.

Characteristic Frequency (%)
Male 79 (45.1)
Maternal education at baseline
 Less than high school 60 (34.3)
 High school 65 (37.1)
 Some college/tech school 43 (24.6)
 College graduate 7 (4.0)
Maternal self-reported race and ethnicity
 Non-Hispanic White 35 (20.0)
 Non-Hispanic Black 109 (62.3)
 Hispanic 29 (16.6)
 Other race 2 (1.1)
Maternal household income at baseline ($), mean (SD) 25,866.74 (25,023.59)
SLC6A4 insertion/deletion frequencies
 “LL” 81 (46.6)
 “LS” 73 (42.0)
 “SS” 16 (9.2)
 “XL” 3 (1.7)
 “XS” 1 (0.6)
BDNF genotype frequencies
 “CC” 109 (62.6)
 “CT” 59 (33.9)
 “TT” 6 (3.5)
COMT genotype frequencies
 “AA” 21 (12.0)
 “AG” 71 (40.6)
 “GG” 83 (47.4)
Open in a new tab

Note. N = 175. Data are provided as frequency (%) except where noted.

Table 2.

Analysis of Variance (ANOVA) Table for Composite Cognitive Score Model.

Dependent Variable df Sums of Squares Mean Square F P Adjusted R 2
Composite scores 5 124,287.0 24,857.40 96.36 <.0001 .17
 Error 2411 621,872.1 257.97
 Corrected total 2416 746,259.1
Open in a new tab

Two genotypes were statistically significant predictors: BDNF “CC” and SLC6A4 insertion/deletion forms “XL.” Both genotypes conveyed a positive effect on the overall model. SLC6A4 “XL” genotype was associated with cognitive composite scores of .88 standard deviations higher compared to carriers of at least one “S” allele; however, it should be noted that the “X” variant is a rare form with minimal representation in this sample; thus, this estimate should be interpreted with caution. Children homozygous for BDNF “C” allele scored .62 standard deviations higher than children with one or more “T” alleles. “T” allele carriers represented 37.36% of the sample; thus, the estimates for this variable are likely robust. COMT did not significantly predict cognitive composite scores. We have presented coefficients and predictor statistical tests for the composite model and coefficients only for individual cognitive tests in Table 3.

Table 3.

Regression Coefficients and Statistical Tests.

Predictor Composite WJ9 WJ10 WISC PPVT
β SE 95% CI β SE β SE β SE β SE
Intercept 19.78 1.80 −0.49 to 0.67 .09 .29 −.03 .15 12.57 0.59 101.34 9.05
BDNF “CC” 0.62* 0.28 0.06 to 1.18 .62 .28 .21 .13 1.71 0.73 13.19 6.24
SLC6A4 “LL” −0.34 0.28 −0.91 to 0.22 −.34 .28 −.13 .11 −1.28 0.96 −7.71 9.08
SLC6A4 “XL” 0.88* 0.42 0.02 to 1.74 .88 .42 .07 .46 0.40 6.63 −4.97 5.56
COMT “GG” 0.29 0.27 −0.27 to 0.85 .29 .27 .07 .12 −0.39 0.93 9.31 8.79
Male −0.33 0.23 −0.79 to 0.13 −.33 .23 −.05 .10 −0.92 0.96 2.62 8.96
Open in a new tab

Note. t Tests were performed only on the composite-measure statistics. No statistical significance test results are reported for the individual cognitive test results in order to contain Type I error. CI = confidence interval; PPVT = Peabody Picture Vocabulary test; WISC = Wechsler Intelligence Scale for Children; WJ9 = Woodcock Johnson Passage Comprehension (subtest 9); WJ10 = Woodcock Johnson Applied Problems (subtest 10).

*p < .05.

In our analysis of individual cognitive tests, children with BDNF “CC” genotype consistently outperformed “T” allele carriers on all our secondary tests, in terms of direction of relationship, with moderate effects compared to the total possible range of scores. For instance, children homozygous for BDNF “C” scored 13.19 points higher on average than “T” allele carriers on the PPVT, an amount equal to 7% of the total possible score. Children with BDNF “CC” genotype demonstrated better scores in general cognition, receptive language, verbal ability, reading comprehension, and symbolic language compared to “T” carriers and outperformed “T” carriers on the composite measure by approximately 13 points out of a total possible score of 185.

Discussion

These exploratory findings from secondary analyses of existing data provide preliminary support for the roles of the copy number variant of SLC6A4 and SNP rs4074134 of BDNF on the development of cognition in children with LBW, with relatively large effect sizes compared to typical findings in genomic research of complex developmental processes in general populations. These candidate genes are biologically plausible agents for affecting the development of general and domain-specific cognition and may interact with the physiologic fragility of children with LBW to produce stronger than typical effects. Further research is required to replicate these findings in a sample where the etiologies of LBW (e.g., preterm birth or fetal growth restriction) and additional genotype differences are known and controlled for in analysis.

The SLC6A4 copy number variant has been studied extensively in relation to neuropsychological disorders but has not been examined in relation to cognitive development in children born preterm or at LBW. The few studies that have examined SLC6A4 in relation to cognition have generally excluded or not controlled for LBW or preterm birth. For example, Maddox et al. (2017) examined the effect of SLC6A4 genotype on cognitive endophenotypes in adults and found that those with low-performing variants of the gene had an advantage in reflexive learning but a disadvantage in reflective learning compared to those with high-performing gene variants. Briefly, reflexive learning involves learning preprogrammed actions to attain immediate rewards (i.e., Pavlovian responses), whereas reflective learning uses working memory and executive function to build, test, and maintain explicit rules. Similarly, Kruijt, Putman, and Van der Does (2014) found evidence of differential attention to negative affective cues in depressed individuals based on S or L(A) genotype. Babineau et al. (2015) explicitly excluded individuals with developmental delay or preterm birth in a study that demonstrated differential self-regulation in children based on SLC6A4 genotype, with an interaction effect from exposure to maternal prenatal depression. Exposure to maternal factors predisposing children to LBW may contribute to differential outcomes in cognition and requires further study.

In our findings, children with one or more “T” alleles of BDNF rs4074134 demonstrated lower test scores than children homozygous for “C” alleles, indicating that “T” allele carriers may be at increased risk of poor cognitive development after LBW. The importance of BDNF to brain development has been well described. Specifically, the BDNF protein is necessary for the survival and proliferation of striatal neurons and the pathfinding and interconnectivity of striatal neurons in the developing brain (Baydyuk & Xu, 2014). BDNF rs4074134 SNP has not been tested in relation to cognitive development, to our knowledge. However, this SNP has been associated with obesity and was included in the Fragile Families study due to this association. Obesity has also been associated with deficits in general cognition, executive function, short-term memory, and language processing in children and adults (Smith, Hay, Campbell, & Trollor, 2011). Thus, it is possible that rs4074134 plays a dual role in the development of general cognition and language processing as well as in the development of obesity. This novel finding opens an exciting avenue for future investigation, and the large effect sizes may indicate that this variant underpins clinically significant developmental processes in this high-risk population. However, we were unable to control for additional variation in BDNF in this analysis due to the limitations of this secondary analysis of existing data. Future research should focus on confirming these findings in an independent sample by including other potentially confounding variants of the BDNF gene, including Val66Met, which has been more extensively studied in relation to cognition. Additionally, research focused on epigenetic modification and expression of this SNP may be useful for determining whether a biomarker for cognitive risk can be developed.

While the candidate genes we chose were not all statistically significant in this analysis, the null findings should be interpreted with caution. Several limitations of this analysis likely artificially reduced the magnitude of findings. Specifically, SLC6A4 has been described as triallelic due to a commonly occurring SNP at location rs2553 (A to G; OMIM, 2016b). The three alleles S, L(A), and L(G) have been described extensively in recent literature, but the Fragile Families study did not measure rs2553. As L(G) is considered a low-expression allele similar in function to the S allele, the combined L group in the Fragile Families data may underestimate differences between the S and L groups. While this underestimation likely biased our findings toward a smaller effect size and may have resulted in the nonsignificant findings in the “LL” group, this assumption should be empirically tested. In addition, the more numerous variants of SLC6A4 likely reduced the power of our analysis to detect changes in specific allelic groups. However, the significant finding regarding the “X” variant indicates a potentially important contribution to cognition in children with LBW from SLC6A4; additional research in a larger sample is needed. Based on the relatively large effect sizes that we found compared to overall test range, power may have played a role in the lack of statistical significance for COMT; a larger sample size may be able to more adequately characterize the contributions of these genes and potential gene–gene interactions.

Finally, due to legal and ethical restrictions on the Fragile Families data at the time we were planning this analysis and it was approved by the IRB, we were unable to control for gestational age and our measure of LBW is by maternal report. Gestational age and continuous birth weight in the Fragile Families data is available only in the medical records data. Fragile Families does not allow medical records data to be combined with genomic data for analyses to limit the potential of participant reidentification. While the Fragile Families investigators have reported that maternal report in this sample overlapped birth weights obtained through medical records in 98% of cases, the lack of ability to control for gestational age or continuous birth weight is a limitation to this analysis; thus, we were unable to interpret findings in relation to LBW etiology (e.g., preterm birth, fetal growth restriction). The scientific literature on LBW and preterm infants is highly confounded due in part to historical definition changes and the significant degree of overlap between these two populations (about 70% of children with LBW are also preterm). We were also not able to control for the receipt of early intervention. However, the receipt of early intervention was likely predicated on birth weight and/or gestational age and perhaps on geographic location (given the inconsistent standards for referral across U.S. geographic regions) rather than genotype. Future research into the genetic contribution of SLC6A4 and BDNF on cognitive development after LBW should focus on clinically confirmed LBW and preterm infants, with additional controls included in models to enable more precision in interpretation.

Ethical Considerations

Caution must be taken in the reporting and study of genetic predictors of development to ensure that findings are not misinterpreted as actionable in prenatal or preimplantation screening. Because neurodevelopment is a complex process and not a clearly defined life-limiting or debilitating disorder, gene targets for cognitive deficits are not amenable to direct intervention due to ethical and practical concerns (National Academies of Sciences, Engineering, & Medicine, 2017). Moreover, cognitive ability is the product of a complex interaction between genetic substrate and environment. Thus, while genetic predisposition to cognitive problems may not be directly actionable, developmental risks may be modified in other ways. Complex disease and developmental phenotypes (e.g., cardiovascular disease or cognition) are typically comprised of both modifiable and nonmodifiable risks. Early identification of nonmodifiable risks often provides advanced warning and enhances our ability to allocate resources and to engage individuals, families, and societies in modifying the risks that are amenable to intervention to reduce the overall burden of disease. Penetrance is incomplete and the contributions of other genes, epigenetic modifications, and supportive environments may overcome the risks posed by one or even a cluster of risk genotypes. In essence, predisposition is not destiny. Thus, inclusion of genetic predictors of risk for cognitive impairment following adverse birth outcomes offers great opportunity to inform and prompt action before the onset of potentially irreversible developmental delay.

Conclusions

This analysis contributes novel findings related to the protective effect of BDNF rs4074134 on the development of cognition in children with LBW. Further, this study addresses a gap in the literature regarding the contributions of genetic variation to cognitive development in an LBW population at high risk for cognitive deficits across the lifespan. While genetic contributions to complex developmental processes such as cognition are not directly amenable to intervention, the ability to detect risk and protective factors of the magnitude described in this analysis at birth may someday provide clinicians with clearer guidance on making referrals to resource-limited early intervention programs. Further research is needed to confirm these preliminary findings, to address several limitations inherent in this study design, and to continue to build on this important body of work.

Acknowledgments

We would like to thank the researchers and staff of the Fragile Families and Child Wellbeing Study and the families who so generously participated over the course of more than 9 years of data collection.

Authors’ Note: The contents are the responsibility of the authors and do not necessarily represent the official views of the NIH.

Author Contributions: Lisa M. Blair contributed to conception and design, acquisition, analysis, and interpretation; drafted manuscript; gave final approval; agrees to be accountable for all aspects of work ensuring integrity and accuracy. Rita H. Pickler contributed to conception, design, acquisition, and interpretation; critically revised manuscript; gave final approval; agrees to be accountable for all aspects of work ensuring integrity and accuracy. P. Cristian Gugiu contributed to design and analysis; critically revised manuscript; gave final approval; agrees to be accountable for all aspects of work ensuring integrity and accuracy. Jodi L. Ford contributed to design, acquisition, and interpretation; critically revised manuscript; gave final approval; agrees to be accountable for all aspects of work ensuring integrity and accuracy. Cindy Munro contributed to conception, design, and interpretation; critically revised manuscript; gave final approval; agrees to be accountable for all aspects of work ensuring integrity and accuracy. Cindy M. Anderson contributed to conception, design, acquisition, and interpretation; critically revised manuscript; gave final approval; agrees to be accountable for all aspects of work ensuring integrity and accuracy.

Declaration of Conflicting Interests: The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was made possible by funding through two Ruth L. Kirschstein National Research Service Awards from the National Institutes of Health (NIH), National Institute of Nursing Research (T32NR014225 [Pickler & Melnyk, MPI] and F31NR016623 [Blair, PI]). The Fragile Families and Child Wellbeing Study was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) of the NIH under award numbers R01HD36916, R01HD39135, and R01HD40421, as well as a consortium of private foundations.

ORCID iD: Lisa M. Blair Inline graphic https://orcid.org/0000-0002-6169-5876

References

  1. Aarnoudse-Moens C. S. H., Weisglas-Kuperus N., Goudoever J. B. V., Oosterlaan J. (2009). Meta-analysis of neurobehavioral outcomes in very preterm and/or very low birth weight children. Pediatrics, 124, 717–728. doi:10.1542/peds.2008-2816 [DOI] [PubMed] [Google Scholar]
  2. AceView. (n.d.-a). Catechol-o-methyltransferase: COMT. Retrieved from https://www.ncbi.nlm.nih.gov/ieb/research/acembly/av.cgi?db=human&term=comt&submit=Go
  3. AceView. (n.d.-b). Homo sapiens gene BDNF, encoding brain-derived neurotrophic factor. Retrieved from https://www.ncbi.nlm.nih.gov/ieb/research/acembly/av.cgi?db=human&c=Gene&l=BDNF
  4. Alsobrook J. P., Zohar A. H., Leboyer M., Chabane N., Ebstein R. P., Pauls D. L. (2002). Association between the COMT locus and obsessive-compulsive disorder in females but not males. American Journal of Medical Genetics, 114, 116–120. [DOI] [PubMed] [Google Scholar]
  5. Ansorge M. S., Zhou M., Lira A., Hen R., Gingrich J. A. (2004). Early-life blockade of the 5-HT transporter alters emotional behavior in adult mice. Science, 306, 879–881. doi:10.1126/science.1101678 [DOI] [PubMed] [Google Scholar]
  6. Association of Maternal & Child Health Programs. (2013). Partnering to promote follow-up care for premature infants. Retrieved from http://www.amchp.org/programsandtopics/womens-health/resources/Documents/Follow-Up-Care-for-Premature-Infants-FINAL-Mar2013.pdf
  7. Babineau V., Green C. G., Jolicoeur-Martineau A., Bouvette-Turcot A. A., Minde K., Sassi R.,…MAVAN project. (2015). Prenatal depression and 5-HTTLPR interact to predict dysregulation from 3 to 36 months—A differential susceptibility model. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 56, 21–29. doi:10.1111/jcpp.12246 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bailey B. A., Cole L. K. J. (2009). Rurality and birth outcomes: Findings from Southern Appalachia and the potential role of pregnancy smoking. The Journal of Rural Health, 25, 141–149. doi:10.1111/j.1748-0361.2009.00210.x [DOI] [PubMed] [Google Scholar]
  9. Baydyuk M., Xu B. (2014). BDNF signaling and survival of striatal neurons. Frontiers in Cellular Neuroscience, 8, 254 doi:10.3389/fncel.2014.00254 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Bendheim-Thoman Center for Research on Child Wellbeing, and Department of Molecular Biology. (2016). Fragile families genetic component/DNA restricted use appendage. Retrieved from http://fragilefamilies.princeton.edu/sites/fragilefamilies/files/ff_gen_9y_pub1.pdf
  11. Blair L. M., Pickler R. H., Anderson C. (2016). Integrative review of genetic factors influencing neurodevelopmental outcomes in preterm infants. Biological Research for Nursing, 18, 127–137. doi:10.1177/1099800415605379 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Center for Research on Child Wellbeing. (2015). Fragile families and child wellbeing study. Retrieved from http://www.fragilefamilies.princeton.edu
  13. Esaki T., Cook M., Shimoji K., Murphy D. L., Sokoloff L., Holmes A. (2005). Developmental disruption of serotonin transporter function impairs cerebral responses to whisker stimulation in mice. Proceedings of the National Academy of Sciences of the United States of America, 102, 5582–5587. doi:10.1073/pnas.0501509102 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gogos J. A., Morgan M., Luine V., Santha M., Ogawa S., Pfaff D., Karayiorgou M. (1998). Catechol-O-methyltransferase-deficient mice exhibit sexually dimorphic changes in catecholamine levels and behavior. Proceedings of the National Academy of Sciences of the United States of America, 95, 9991–9996. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hamilton B. E., Martin J. A., Osterman M. J. K. (2018). Births: Provisional data from 2017. Retrieved from https://www.cdc.gov/nchs/data/vsrr/report004.pdf
  16. Herrera B. M., Keildson S., Lindgren C. M. (2011). Genetics and epigenetics of obesity. Maturitas, 69, 41–49. doi:10.1016/j.maturitas.2011.02.018 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hilderman C. G. E., Harris S. R. (2014). Early intervention post-hospital discharge for infants born preterm. Physical Therapy, 94, 1211–1219. doi:10.2522/ptj.20130392 [DOI] [PubMed] [Google Scholar]
  18. Kroll J., Karolis V., Brittain P. J., Tseng C. E. J., Froudist-Walsh S., Murray R. M., Nosarti C. (2017). Real-life impact of executive function impairments in adults who were born very preterm. Journal of the International Neuropsychological Society, 23, 381–389. doi:10.1017/S1355617717000169 [DOI] [PubMed] [Google Scholar]
  19. Kruijt A. W., Putman P., Van der Does W. (2014). The 5-HTTLPR polymorphism, early and recent life stress, and cognitive endophenotypes of depression. Cognition & Emotion, 28, 1149–1163. doi:10.1080/02699931.2013.873018 [DOI] [PubMed] [Google Scholar]
  20. Lotrich F. E., Pollock B. G., Ferrell R. E. (2003). Serotonin transporter promoter polymorphism in African Americans: Allele frequencies and implications for treatment. American Journal of Pharmacogenomics, 3, 145–147. [DOI] [PubMed] [Google Scholar]
  21. Maddox W. T., Gorlick M. A., Koslov S., McGeary J. E., Knopik V. S., Beevers C. G. (2017). Serotonin transporter genetic variation is differentially associated with reflexive-and reflective-optimal learning. Cerebral Cortex, 27, 1182–1192. doi:10.1093/cercor/bhv309 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. National Academies of Sciences, Engineering, and Medicine. (2017). Human genome editing: Science, ethics, and governance. Washington, DC: National Academies Press; doi:10.17226/24623 [PubMed] [Google Scholar]
  23. OMIM. (2016. a). Catechol-o-methyltransferase; COMT. Retrieved from https://www.omim.org/entry/116790
  24. OMIM. (2016. b). Solute carrier family 6 (neurotransmitter transporter, serotonin), member 4, SLC6A4. Retrieved from https://www.omim.org/entry/182138?search=slc6a4&highlight=slc6a4
  25. OMIM. (2017). Brain derived neurotrophic factor: BDNF. Retrieved from https://www.omim.org/entry/113505?search=bdnf&highlight=bdnf
  26. Page D. T., Kuti O. J., Prestia C., Sur M. (2009). Haploinsufficiency for PTEN and serotonin transporter cooperatively influences brain size and social behavior. Proceedings of the National Academy of Sciences of the United States of America, 106, 1989–1994. doi:10.1073/pnas.0804428106 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Peabody Picture Vocabulary Test, Fourth Edition. (n.d). Retrieved from http://www.pearsonclinical.com/language/products/100000501/peabody-picture-vocabulary-test-fourth-edition-ppvt-4.html#tab-resources
  28. Schrank F. A., McGrew K. S., Woodcock R. W. (2001). Woodcock-Johnson III: Assessment service bulletin number 2. Retrieved from http://www.riversidepublishing.com/products/wjIIIComplete/pdf/WJIII_ASB2.pdf
  29. Smith E., Hay P., Campbell L., Trollor J. N. (2011). A review of the association between obesity and cognitive function across the lifespan: Implications for novel approaches to prevention and treatment. Obesity Reviews, 12, 740–755. doi:10.1111/j.1467-789X.2011.00920.x [DOI] [PubMed] [Google Scholar]
  30. Tobacco & Genetics Consortium. (2010). Genome-wide meta-analyses identify multiple loci associated with smoking behavior. Nature Genetics, 42, 441–447. doi:10.1038/ng.571 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Vijayendran M., Cutrona C., Beach S. R. H., Brody G. H., Russell D., Philibert R. A. (2012). The relationship of the serotonin transporter (SLC6A4) extra long variant to gene expression in an African American sample. American Journal of Medical Genetics, 159B, 611–612. doi:10.1002/ajmg.b.32054 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Williams P. E., Weiss L. G., Rolfhus E. L. (2003). WISC-IV technical report #2: Psychometric properties. Retrieved from https://www.pearsonclinical.com.au/files/TechReport2WISCIVPsychometrics.pdf
  33. Zhao J., Bradfield J. P., Li M., Wang K., Zhang H., Kim C. E.…Grant S. F. (2009). The role of obesity-associated loci identified in genome wide association studies in the determination of pediatric BMI. Obesity (Silver Spring), 17, 2254–2257. doi:10.1038/oby.2009.159 [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Biological Research for Nursing are provided here courtesy of SAGE Publications

RESOURCES