Household socioeconomic status relates to specific hippocampal subfield volumes across development

Abstract

The hippocampus is composed of cytoarchitecturally-distinct subfields that support specific memory functions. Variations in total hippocampal volume across development have been linked to socioeconomic status (SES), a proxy for access to material resources, medical care, and quality education. High childhood household SES is associated with greater cognitive abilities in adulthood. Currently, it is not known whether household SES differentially impacts specific hippocampal subfield volumes. We assessed susceptibility of subfields to variations in household SES across development in a sample of 167 typically developing 5- to 25-year-olds. Bilateral cornu ammonis (CA) 1–2, combined CA3-dentate gyrus (DG), and subiculum (Sub) volumes were measured by highly-reliable manual segmentation of high-resolution T2-weighted images and adjusted for intracranial volume. A summary component score of SES measures (paternal education, maternal education, and income-to-needs ratio) was used to examine variability in volumes across ages. We did not identify age-related differences in any of the regional volumes, nor did age modify SES-related effects. Controlling for age, larger volumes of CA3-DG and CA1–2 were associated with lower SES, while Sub volume was not. Overall, these findings support the specific impact of SES on CA3-DG and CA1–2 and highlight the importance of considering environmental influences on hippocampal subfield development.

The hippocampus supports memory (e.g, Squire & Wixted, 2011), language comprehension (e.g., Duff & Brown-Schmidt, 2017), and social and emotional (Immordino-Yang & Singh, 2013) functions important for daily life (Olsen et al., 2012). The hippocampus is comprised of subfields distinct in cytoarchitecture, function, and developmental trajectory: dentate gyrus (DG), cornu ammonis 1–3 (CA1–3), and subiculum (Sub) (Duvernoy et al., 2005; Insausti & Amaral, 2012). Evidence from nonhuman primates suggests different rates of structural development across subfields, with the most protracted development of DG and CA3 subfields (Lavenex & Banta Lavenex, 2013; Seress, 2001). These findings led to hypotheses postulating a link between hippocampal subfield development and the cumulative gains in memory functions between childhood and adulthood. In the extant developmental human neuroimaging studies, the link between hippocampal subfield volumes and age appear to vary in magnitude and even direction (Canada et al., 2021; Daugherty et al., 2017; Lee et al., 2014; Riggins et al., 2018; Schlichting et al., 2017). Some variations in this body of literature stem from methodological differences, such as variability in the age ranges studied, method used to measure volumes (e.g., along the long axis versus within hippocampal body), and cross-sectional versus longitudinal approaches. However, when considered cumulatively across the lifespan, findings suggest increasing volume from early childhood to young adulthood, followed by loss of volume later in life (Bussy et al., 2021; Canada et al., 2021; Daugherty et al., 2016; Lee et al., 2014; Schlichting et al., 2017).

Importantly, the most consistent finding across studies of hippocampal subfields in developmental samples is a large between-subject variability that implies strong influence of myriad of factors on hippocampal subfield volumes beyond age alone. Environmental factors modify hippocampal development from infancy into adulthood (Tottenham & Sheridan, 2009), accounting for 10–30% of variability in overall brain development (Lenroot & Giedd, 2008). Taken together, linking environmental factors to individual differences in hippocampal subfield volumes may help characterize hippocampal development, and given the crucial role of this structure in learning and memory, will help improve understanding of individual differences in learning and memory across the lifespan.

Socioeconomic status (SES) is a proxy for multiple environmental factors and individual household variables that likely influence brain and cognitive development. Higher childhood SES is predictive of lifelong outcomes, such as higher cognitive abilities and lower incidence of neurodevelopmental disorders, and this link is partly explained by SES-related vulnerability of the hippocampus (Čermaková et al., 2020; Whittle et al., 2014).

Household SES has been shown to correlate with total hippocampal volume across development (Hanson et al., 2011; Yu et al., 2018). For example, lower household income correlates with smaller total hippocampal volume and blunted stress reactivity (Raffington et al., 2019). Less is known about possible differential vulnerability of specific hippocampal subfields to factors associated with SES. In animal studies, increased vulnerability of CA1, CA3, and DG subfields to stress was evinced in reduced dendritic branching, less synaptic plasticity, and spine density (McEwen et al., 2016). In humans, larger DG-CA3 volumes correlate with lower hair cortisol levels, a metric of stress, in 6-to 7-year-olds (Keresztes et al., 2020). Higher parental education also correlates with lower hair cortisol levels in 5-to 9-year-olds (Merz et al., 2019). Further, larger CA1, DG-CA4, and Sub hippocampal subfield volumes correlate with higher SES in 9- to 13-year-olds (Botdorf et al., 2022). Thus, childhood appears to be a period when hippocampal structures are vulnerable to factors associated with household SES, providing a window onto mechanisms of cognitive change associated with early-life SES effects on cognitive development.

Here, we test the unique susceptibility of CA1–2 and CA3-DG hippocampal subfield volumes to variation in SES over and above chronological age effects. Based on the evidence reviewed above we hypothesized larger volumes of CA1–2 and CA3-DG hippocampal subfields would relate to higher household SES, and a weak correlation between Sub and household SES. One hundred sixty-seven 5-to-25-year-olds (M_age = 13.44, SD_age = 5.40, 85 females, 65% White, 23% Black, 10% Asian, 2% Other/Not Reported) participated in the present study. The sample included 65 typically developing children (5–12 years), 69 typically developing adolescents (13–17 years), and 33 young adults (18–25 years). Participants were recruited from the Metro Detroit area as part of a larger study of cognitive and brain development. Participants were self-reported right-hand dominant, spoke English as a native language, had no reported developmental or neurological disorders, and no history of head trauma. For MRI compatibility and safety, participants had no metallic implants, braces, or permanent retainers. Participants were consented in accord with procedures approved by the University Institutional Review Board, which included parental consent for minors.

The neuroimaging parameters and manual segmentation procedures have been described before (Daugherty et al., 2017; Homayouni et al., 2021) and are described here in brief. Highly-reliable manual segmentation (ICC(2) > 0.85) of hippocampal subfields was implemented on high-resolution (0.4 × 0.4 × 2.0 mm³) proton density-weighted, turbo spin echo (PD-TSE) sequence that was acquired perpendicular to the long axis of the hippocampus. The protocol delineates cornu ammonis 1–2 (CA1–2), CA3-dentate gyrus (CA3-DG), and subiculum (Sub) following a geometric heuristic in the length of the hippocampal body (Figure 1). Hippocampal subfield volumes were adjusted for intracranial volume (ICV) with an analysis of covariance approach (Jack et al., 1989; Raz et al., 2005), after confirming homogeneity of regression slopes by age and sex (all p > .12). Analysis of covariance volume correction approach is commonly used and recommended based on comparisons of different approaches in the literature (e.g., Voevodskaya et al., 2014). Intracranial volume was estimated and extracted using T1-weighted MPRAGE images (0.5 mm × 0.5 mm × 1.0 mm³) using the brain extraction tool (Smith, 2002) in FSL following previously reported procedures (Bender et al., 2013).

Figure 1.

Example of manual demarcation of hippocampal subfields from high-resolution (0.4 × 0.4 × 2 mm³) PD-TSE-T₂-weighted image in the coronal plane. CA3-DG is shown in blue, CA1–2 in red, and Sub in green.

SES as a latent proxy was calculated as a principal component score (varimax rotation) from participants’ maternal and paternal education (ordinal variables), and income-to needs ratio based on median income level and family size relative to federal poverty level (Yu et al., 2018). The three measures loaded on a single component (loadings = 0.54–0.84), accounting for 57% of the shared variance, and the standardized composite score was used in subsequent analysis (see Table 1). To retain subjects with partially missing data for analysis, missing values of maternal education (n =21), paternal education (n = 20), and income-to-needs ratio (n = 13) were replaced with the sample mean. Substantive findings of the analysis remained the same when excluding subjects with missing data, thus results of the analysis with the total sample are reported to prioritize external validity.

Table 1.

Descriptive statistics of the three socioeconomic status measures

SES measure	Mean (SD)	Range	Loading
Income-to-needs ratio	2.87 (1.22)	(0.26, 5.44)	.54
Father’s education	4.97 (1.44)	(1, 7)	.84
Mother’s education	5.55 (1.32)	(2, 7)	.85
Standardized Composite SES score	0.00 (1.00)	(−2.85, 1.72)

Note: SD = Standard deviation.

Multiple linear regression tested age- and SES-related differences in hippocampal subfield volumes within one model, and additionally controlling for sex. Of note, in this sample, household SES was associated with age (r = −0.29, p < .0001) and not ICV (r = 0.12, p > .10), therefore age was maintained as a covariate in further analysis. Analyses were completed for each region separately and Bonferroni correction of significance threshold was made to control type I error rate (α’ = .017).

Independent of SES, variations in CA3-DG (b_age = 1.80, p = .20, η²_p = .010 and b_sex = −24.07, p = .10, η²_p = .017), CA1–2, (b_age = 0.17, p = .89, η²_p = .0001 and b_sex = −4.70, p = .71, η²_p = .001) and Sub volumes (b_age = 1.102 p = .49, η²_p = .003 and b_sex = −21.28, p = .16, η²_p = .012) did not relate to chronological age (Figure 2). Follow up tests using models examining only the relation between chronological age (i.e., excluding SES) while controlling for sex did not identify age effect, supporting no significant age-related differences in hippocampal subfield volumes in this sample.

Figure 2.

Associations between chronological age in years and A) CA3-DG (blue), B): CA1–2 (red), and C) Sub (green) ICV-adjusted volumes, controlling for sex and SES. Adjusted correlations were not statistically significant.

Independent of age, variations in CA3-DG (b_SES = −16.81, p = .0168, η²_p = .035) and CA1–2 (b_SES = −18.26, p = .0168, η²_p = .035) were associated with SES. In contrast, Sub volume was not associated with SES (b_SES = −2.10, p = .79, η²_p = .0004). Specifically, lower composite SES was associated with larger CA3-DG and CA1–2 volumes in this sample (Figure 3). Steiger’s z tests to compare semi-partial correlations of volumes and SES (adjusted for age) between regions revealed no significant differences between CA3-DG and CA1–2 (z = −0.072, p = .94). However, the correlation between SES and Sub volume was smaller than the correlation between SES and either CA3-DG (z = −2.063, p = .039), or CA1–2 (z = −2.059, p = .040). To explore whether relations between hippocampal subfields and SES were modified by age, additional models including the interaction between SES and age (controlling for sex) were tested. No support was found for a moderated effect (p_CA3-DG= 0.65, p_CA1–2= 0.34, and p_Sub= 0.82). Thus, socioeconomic effects on hippocampal subfield volumes are strongest for CA3-DG and CA1–2 and do not differ by age across this period of development.

Figure 3.

Relations between household SES and A) CA3-DG (blue), B): CA1–2 (red), and C) Sub (green) ICV-adjusted volumes controlling for sex and age.

The present study highlights the vulnerability of hippocampal subfield volumes to socioeconomic disparities across development in a cross-sectional study of 5- to 25-year-olds. Volumes of CA1–2 and CA3-DG negatively correlated with household SES, when accounting for age. Findings from animal studies suggest several neurobiological processes, including neurogenesis, that account for changes in brain volumes (e.g., Callaghan & Tottenham, 2016; Hackman et al., 2010; Naninck et al., 2015). Moreover, these processes may be modified by stress, with particular sensitivity in CA and DG subfields (McEwen et al., 2016). In humans, higher stress is reported in individuals living in lower SES environments (Baum et al., 1999). Taken together, our finding that household SES correlated with CA1–2 and CA3-DG volumes may reflect stress-related modification of neurodevelopmental processes. Given the importance of the hippocampus in neurodevelopmental disorders and cognitive development, these findings highlight the importance of considering the influences of SES on hippocampal subfield development.

The direction of the effect identified in this study—larger volumes of CA1–2 and CA3-DG subfields related to lower SES—differed from the direction hypothesized based on the extant evidence (Botdorf et al., 2022; Merz et al., 2019). Our finding may be interpreted in the context of the stress acceleration hypothesis that posits faster brain maturation, particularly in the hippocampus, in low SES individuals due to greater chronic stress (Baum et al., 1999; Callaghan & Tottenham, 2016; Humphreys et al., 2019; Tooley et al., 2021). Research across the lifespan suggests a non-linear developmental trajectory of hippocampal subfields, with increasing volume from childhood to young adulthood, followed by loss of volume with aging (Bussy et al., 2021; Canada et al., 2021; Daugherty et al., 2016; Lee et al., 2014; Schlichting et al., 2017). If indeed hippocampal subfields undergo robust growth through childhood, the larger volumes observed in participants from low SES households from this sample may indicate altered rates of maturation. We did not find age-related differences in the current cross-sectional study; thus, it is not yet possible to link the effects of SES in the household to specific developmental effects in humans. As expanded upon below, additional work is needed to directly pursue this avenue of research.

While our findings present an intriguing direction for future research, further studies are needed to generalize beyond the current sample and study limitations. While the protocol adopted in our study measured hippocampal subfield exclusively from the hippocampal body, the two other studies used a wider range along the long axis of the hippocampal to estimate subfield volumes (Merz et al. 2019; Botdorf et al. 2022). The protocol used in our study was adopted for its strength in terms of validity and reliability (Homayouni et al., 2021), yet it is possible that the effects observed are specific to hippocampal body, with different effects along the longitudinal axis given varying patterns of structural and functional development (Canada et al., 2020; Daugherty et al., 2015; DeMaster & Ghetti, 2013). More broadly, the use of manual segmentation method is a key difference in the approach taken in the current study compared to prior work. For example, both Merz et al. (2019) and Botdorf et al. (2022) separately examined CA2/CA3 from CA1 and DG using the automated Freesurfer parcellation (Iglesias et al., 2015) using low-resolution T1-weighted images. Thus the segmentation method used in the current investigation may yield discrepant findings compared to those based on automated methods that may produce measures with validity concerns (Wisse et al., 2020). Further, it is possible that the effects observed are due to sample specific factors such as the household SES range and the age range examined. Advancing our understanding of how environmental factors impact hippocampal maturation will require using more representative samples geographically and socioeconomically across a wide age range. Additionally, with a cross-sectional design we cannot assess developmental trajectories, or evaluate whether SES influence hippocampal development differently across specific periods as implied by the accelerated maturation hypothesis (Callaghan & Tottenham, 2016). Longitudinal study of participants from early childhood to adulthood, with a representative range of household SES is necessary to assess the accelerated maturation hypothesis.

We also want to acknowledge that SES is a distal proxy of multiple factors and further work is needed to disentangle the influence of specific household SES-related factors such as wealth, access to healthcare, and educational enrichment (Krieger et al., 1997) on brain development. Household SES also strongly correlates with neighborhood-level factors, such as environmental pollution, crime, and undernutrition that may further shape hippocampal development into young adulthood (Evans & English, 2002). Future studies that consider the synergistic interaction of household- and neighborhood-level factors related to SES can provide deeper insight into the mechanistic role SES plays in hippocampal maturation. Nonetheless, the reported findings help guide hypotheses in longitudinal work further assessing the vulnerability of CA1–2 and CA3-DG subfields to SES as a proxy for individuals’ household resources.

Given evidence linking CA1–2 and CA3-DG subfields to specific memory functions, the impact of access to resources that differ by SES provides a potential avenue to explore interventions to promote healthy brain and cognitive development. Overall, these findings highlight the importance of considering the role of environmental influences on hippocampal subfield development.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.