Skip to main content
NCBI home page
Brain, Behavior, & Immunity - Health logoLink to Brain, Behavior, & Immunity - Health
. 2026 Sep 10;57:101353. doi: 10.1016/j.bbih.2026.101353

Proteomic insights into childhood adversity: Exploring biological mechanisms and cognitive outcomes in the CARDIA study

Deborah K Rose a,⁎, Gabrielle N Pfund b, Robin Ortiz c, Sithara Vivek d, Roland J Thorpe Jr a,e, Kristine Yaffe f, Indira C Turney g, Keenan A Walker h, Sarah N Forrester i
PMCID: PMC13613509  PMID: 42799052

Abstract

Adverse childhood experiences (ACEs), including abuse, neglect, and household dysfunction, are among the most consequential adverse social experiences of early life and carry elevated risk of chronic disease and cognitive impairment in adulthood through biological pathways that are poorly understood. We examined the mediation of ACE-related plasma proteins, including immune signaling markers, in the association between childhood adversity and cognition in midlife using data from the Coronary Artery Risk Development in Young Adults (CARDIA) cohort. Cognitive performance across verbal memory, processing speed, executive function, and global cognition was assessed with validated neuropsychological batteries. Among 2003 participants (mean [SD] age: 55 [3.6] years; 56% female; 42% Black), adjusted linear regression analyses showed that 35 proteins were significantly associated with at least one ACE (p < 0.001, a priori discovery threshold). Of these, two proteins were associated with cognition after adjusting for age, sex, race, and years of education, and correction for multiple comparisons (Benjamini-Hochberg false discovery rate [FDR], q < 0.05). Higher MCTS1, a regulator of translation re-initiation and cell cycle progression, was associated with worse processing speed (β = −0.088, q = 0.002), and higher AINX, a neuron-specific intermediate filament protein, was associated with worse executive function (β = −0.069, q = 0.025). Adjusted mediation analyses did not identify any significant pathways that may have explained the effects of ACEs on cognition. These findings suggest that ACEs are associated with alterations in specific circulating proteins, including MCTS1 and AINX, that are independently associated with cognitive performance in midlife. Although mediation analyses did not support these proteins as significant intermediaries association ACEs with cognition after correction for multiple comparisons, their convergent identification across independent analyses suggests they may be worth prioritizing as candidate biomarkers for future study.

Keywords: Childhood adversity, Proteomics, Cognition, Biological embedding, CARDIA

Highlights

  • •

    Childhood adversity associates with distinct plasma protein signatures in midlife.

  • •

    MCTS1 and AINX are the most robust of 12 cognition-associated candidate proteins.

  • •

    No proteomic marker mediated childhood adversity-cognition associations.

  • •

    Proteomic signatures may identify adults at risk for adversity-related cognitive change.

1. Introduction

Adverse childhood experiences (ACEs), including abuse, neglect, and household dysfunction, are pervasive and fundamental disruptors of one's social environment and health across the lifespan (Danese and McEwen, 2011; Nelson et al., 2020; Felitti et al., 1998). ACEs are associated with increased risk for many chronic medical conditions (e.g., cardiovascular disease, diabetes, psychiatric disorders, cognitive impairment), which highlights the capacity of early-life adversity to embed biologically and alter developmental and physiological trajectories well into adulthood (Danese and McEwen, 2011; Daskalakis et al., 2015; Danese and Widom; Danese et al., 2009).

Childhood adversity does not appear to affect cognition uniformly. A systematic review found that specific ACE subtypes, including physical neglect and physical or sexual abuse, differentially influenced distinct cognitive domains such as executive function, attention, working memory, and verbal and visual memory, in non-clinical populations (Rosa et al., 2023a). This domain specificity suggests that different forms of adversity may act through distinct mechanisms. One hypothesized pathway through which ACEs exert lasting effects is via chronic dysregulation of immune processes (Bourassa et al., 2021; Lin et al., 2015; Danese et al., 2007). Social psychoneuroimmunology situates this pathway within a broader bidirectional framework: adverse social experiences alter immune function, and disrupted immune signaling, in turn, affects the social-cognitive capacities (e.g., executive function, processing speed) that people rely on for social perception and everyday interpersonal functioning (Muscatell, 2021). Childhood adversity is a particularly potent social stressor within this framework, given that early immune alterations may compound across development and manifest as impaired social cognition well into midlife (Nusslock and Miller, 2015; Chen et al., 2021).

Prior research, however, has typically relied on canonical inflammatory markers (Danese et al., 2007; Baumeister et al., 2016). A meta-analysis found only a small overall association between childhood stress and inflammation, and no evidence that this association differs by adversity type or specific inflammatory marker (Chiang et al., 2022). This constrains inference such that when only immune proteins are assayed, only immune pathways can be recovered. Thus, immune markers may lack the resolution to distinguish how different forms of adversity affect biology. This limits our understanding of the broader proteomic landscape through which ACEs may influence health, particularly cognitive function. Advances in high-throughput proteomics now allow comprehensive, multiplexed measurement of thousands of circulating proteins, including a substantial complement of immune and inflammatory markers alongside proteins spanning metabolic regulation, cellular stress response, and other interconnected systems (Candia et al., 2022). These non-immune systems point toward other routes from early adversity to cognition: cellular stress response and proteostasis pathways govern neuronal management and recovery from damage over time, metabolic regulation directly influences the energetic and structural resources available to support brain aging, and proteins involved in translational control and cytoskeletal integrity maintain the synaptic and axonal architecture underlying memory and executive function (Beste et al., 2019; Hetz, 2021; Na et al., 2025). This offers an unprecedented opportunity to investigate the biological correlates of ACEs and their downstream effects on cognition. Few studies have applied this approach to sociodemographically diverse, community-based cohorts, and the specific protein-level signatures associated with ACEs (particularly those linked to cognition) are largely unknown. Further, most large-scale population studies have not examined ACE–protein associations in midlife, which is a critical period for both dementia risk detection and intervention.

The Coronary Artery Risk Development in Young Adults (CARDIA) cohort, which recently incorporated an extensive proteomic panel comprising >6600 plasma proteins measured by the SomaScan assay, offers a unique opportunity to address these gaps. Given the biological plausibility and clinical relevance of protein-level alterations as potential mechanistic links between ACEs and cognition, we sought to identify plasma proteins broadly associated with different ACEs reported by older and midlife adults and to examine their associations with cognition. Specifically, our study assessed: (1) proteome-wide associations between five ACEs (emotional neglect, emotional abuse, physical abuse, household substance use, and household structure and supervision) and plasma protein levels; (2) associations between these ACEs and cognition; (3) associations between these ACE-related proteins and cognitive performance; and (4) the mediating role of these ACE-related proteins in ACE–cognition associations. We hypothesized that higher ACE exposure would associate with multiple biological proteins beyond immune markers, and that these ACE-related proteins would mediate effects on cognitive performance. Because childhood adversity and its biological correlates may be experienced differently across demographic groups, we also examined the variation of these associations by age, race, and sex.

2. Methods

2.1. Study design and participants

We included participants from the CARDIA study, an ongoing multicenter, population-based prospective cohort initiated in 1985 that investigates the evolution of cardiovascular risk from young adulthood into midlife (Friedman et al., 1988). Participants were aged 18-30 at the study's baseline (1985-86). This longitudinal cohort includes 5115 Black and White individuals recruited from four US cities: Birmingham, AL; Chicago, IL; Oakland, CA; and Minneapolis, MN (Friedman et al., 1988). This study received IRB approval from the four participating field centers (University of Alabama at Birmingham, Northwestern University, Kaiser Permanente Northern California, and University of Minnesota) and the coordinating center at University of Alabama at Birmingham. All participants provided informed consent before study enrollment.

2.2. Adverse childhood experiences (ACEs)

ACEs were assessed at Year 15 (2000–2001), the only wave at which the Childhood Family Environment questionnaire was administered, which captures five dimensions of adversity experienced before age 18 years. This questionnaire contains the complete set of childhood adversity content available in CARDIA; no additional items were excluded from this analysis. This same instrument has been used in several prior CARDIA analyses associating childhood adversity with cardiovascular health (Loucks et al., 2011, 2013; Pierce et al., 2020; Ortiz et al., 2024). We examined four standard ACE domains (emotional neglect, emotional abuse, physical abuse, and household substance use) (Felitti et al., 1998; Na et al., 2025) and added household structure and supervision to capture instability not reflected in traditional ACE frameworks. Items were rated on a four-point Likert scale: 1 (rarely/none of the time), 2 (some/little of the time), 3 (occasionally/moderate amount of the time), or 4 (most/all of the time). Each ACE domain was examined individually rather than as a composite score in order to identify the association of distinct forms of adversity with distinguishable proteomic and cognitive signatures.

Emotional neglect averaged two reverse-coded items assessing how often a parent or other adult (1) made the respondent feel loved, supported, and cared for and (2) expressed physical affection. Emotional abuse measured frequency of being sworn at, insulted, or threatened by a household adult, and physical abuse assessed being pushed, grabbed, or hit hard enough to leave marks or cause injury. Household substance use was assessed by asking if the respondent lived with anyone with substance use disorder (e.g., alcohol, street drugs). Household structure and supervision averaged two reverse-coded items: (1) how well-organized and well-managed the household was, and (2) how much the family knew what the respondent was up to as a child. Internal consistency of the emotional neglect and household structure and supervision domains was evaluated using Cronbach's alpha, which indicated good reliability (α = 0.91 and α = 0.85, respectively). All ACEs were analyzed as continuous variables, and any reported exposure was coded as present. Higher values indicated greater adversity.

2.3. Cognitive assessment

Cognitive functioning was assessed at Year 30 (2015–2016) using neuropsychological measures targeting four cognitive domains: verbal memory, processing speed, executive function, and global cognition (Lezak et al., 2012). Verbal memory was measured by the Rey Auditory Verbal Learning Test, which evaluates the ability to encode, consolidate, and retrieve verbal information (Schmidt, 1996). Processing speed was measured using the Digit Symbol Substitution Test from the Wechsler Adult Intelligence Scale (Wechsler, 1955, 1987). Executive function was measured using the Stroop Color and Word Test interference condition to assess the ability to inhibit automatic responses and engage higher-order control processes. Participants were asked to name the ink color when color–word pairs were incongruent (e.g., word “red” in blue ink). Global cognition was assessed with the Montreal Cognitive Assessment (MoCA), a standardized and widely validated instrument for detecting mild cognitive impairment and early dementia (Freitas et al., 2013; Nasreddine et al., 2005). This 30-point instrument assesses visuospatial/executive function, naming, attention, language, abstraction, delayed recall, and orientation (Freitas et al., 2013; Nasreddine et al., 2005).

All scores were standardized to z-scores, and higher z-scores indicated better performance. Cognitive domains were examined individually rather than as a composite score, given evidence that specific forms of childhood adversity are differentially associated with specific cognitive domains rather than uniformly with global cognitive function (Rosa et al., 2023b). The Rey Auditory Verbal Learning Test, Digit Symbol Substitution Test, and Stroop Color and Word Test were administered at both Year 25 and Year 30, whereas the Montreal Cognitive Assessment was administered only at Year 30. Year 30 measures for all four cognitive domains was used in this analysis because it is the only wave 1) at which global cognition was available, 2) following the Year 25 proteomic assessment (preserving the intended ACE → protein → cognition temporal ordering), and 3) that maximizes the interval since ACE exposure.

2.4. Proteomic biomarkers

Plasma samples were collected at Year 25 (2010–2011), the only CARDIA exam at which the SomaScan proteomic panel was administered. Samples had been analyzed using the SomaScan Version 4.1 platform (SomaLogic Inc.), which quantifies 7335 SOMAmer reagents targeting 6609 unique human proteins. Raw relative fluorescence units were provided by SomaLogic after a multi-step quality-control pipeline (Choi et al., 2024). The median intra-assay coefficient of variation across all aptamers was ∼5%, consistent with high assay reproducibility. Protein expression values were log-transformed to normalize right-skewed distributions prior to statistical analysis. Samples with missing data or flagged as poor quality were excluded from analyses.

2.5. Covariates

Demographic covariates were selected a priori and included: age (continuous, years), sex (male/female), race (Black/White), and educational attainment (years). Education level, an important socioeconomic covariate given its association with both ACEs (Lam et al., 2024; Pan et al., 2019) and later-life cognitive outcomes (Sharp and Gatz, 2011; Yan et al., 2006), was assessed at Exam 5 (Year 10, 1995–96), by which point most participants (then aged 28–40) had completed formal schooling. Age, sex, and race came from the Year 30 visit for concurrent exposure. Sex and race were self-reported at study enrollment. Age, sex, and race were included because each is independently associated with circulating protein abundance and with midlife cognitive performance (Manly et al., 1998; McCarrey et al., 2016; Lehallier et al., 2019; Koprulu et al., 2025).

2.6. Statistical analysis

R version 4.5.0 was used for data cleaning, descriptive statistics, visualization, and four-step analyses. All four steps were adjusted for age, sex, race, and years of education. To account for multiple comparisons within each analytic step, we applied Benjamini-Hochberg false discovery rate (FDR) correction (q < 0.05) independently to Steps 2 through 4. Step 1, an exploratory proteome-wide discovery screen, is treated separately given its substantially larger number of comparisons. (1) Multiple linear regression examined the associations between ACEs and >6600 log-transformed standardized plasma proteins (n = 2003). Given the exploratory, discovery-oriented aims of this analysis, we applied a fixed threshold of p < 0.001 as an a priori feature-selection criterion. This threshold is two orders of magnitude more stringent than a conventional α = 0.05. No ACE–protein association met Bonferroni or Benjamini–Hochberg false discovery rate correction across the full panel of >6600 proteins; the p < 0.001 candidates identified here therefore are a screening set requiring independent replication. (2) Candidate ACE-related proteins were assessed in association with standardized cognitive z-scores (n = 1811) via linear regression. Given 140 tests at this stage (35 candidate proteins × 4 cognitive domains), we applied FDR correction (q < 0.05). (3) We then tested associations between ACEs and cognitive z-scores (n = 1587); given 20 tests (5 ACE domains × 4 cognitive domains), we applied FDR correction (q < 0.05). (4) Finally, we evaluated if ACE-related proteomic markers mediated ACE–cognition associations (n = 1587), testing all 35 candidate proteins identified in Step 1 across each combination of ACE and cognitive outcome (700 pathways total: 35 proteins × 5 ACE domains × 4 cognitive domains). For each pathway, we estimated the average causal mediation effect (ACME), average direct effect, total effect, and proportion mediated using nonparametric bootstrapping (10,000 simulations), adjusting for age, sex, race, and education (Tingley et al., 2014). FDR correction (q < 0.05) was applied across the 700 tests.

These assessments were in temporal order, which reduces concern about reverse causation but does not establish causality: ACEs (analytic baseline at Year 15) preceded proteomics (Year 25), which preceded cognition (Year 30). For steps 1–3, associations were stratified by age (older midlife [≥55 years] vs. younger midlife [<55 years]; 55 was the mean age of the analytic sample), race (Black vs. White), and sex (female vs. male). These stratified analyses were exploratory and undertaken because the prevalence and reporting of ACEs, circulating proteomic signatures, and midlife cognitive performance are each known to differ across these groups (Manly et al., 1998; McCarrey et al., 2016; Lehallier et al., 2019; Koprulu et al., 2025; Mersky et al., 2021). Within each stratified analysis, we applied FDR correction (q < 0.05) independently to each test family (ACE-protein: 1050 tests; protein-cognition: 840 tests; ACE-cognition: 120 tests). Mediation analyses were conducted using the mediation package in R.

3. Results

3.1. Demographics and clinical characteristics

Analytic samples included 1587–2003 participants from the CARDIA Year 30 exam with complete data for relevant domains (Table 1). The age range was 47–64 years (mean: 55.2; SD: 3.6), and the average years of education were 14.8 (SD: 2.6). Across samples, women comprised 56% and men 44% of participants. Racial composition was 42% Black and 57% White. The prevalence of individual ACEs was 43.8% for emotional neglect (15.1% moderate or above), 40.6% for emotional abuse (17.9% moderate or above), 19.0% for physical abuse (7.0% moderate or above), 28.5% for household substance use (20.0% moderate or above), and 40.3% for household structure and supervision (12.5% moderate or above); any exposure reflects scores ≥2 and "moderate or above" reflects scores ≥3 on the four-point scale described above. 56.6% of participants scored below the conventional MoCA cutoff of 26. This reflects the raw MoCA total score without the standard one-point education correction.

Table 1.

Demographic and clinical characteristics in the CARDIA cohort.

Characteristic Analytic sample for relationship between ACEs and proteomics (n = 2003) Analytic sample for relationship between proteomics and cognition (n = 1811) Analytic sample for relationship between ACEs and cognition (n = 1587)
Demographics M (SD) M (SD) M (SD)
Age (years) 55.3 (3.60) 55.2 (3.59) 55.3 (3.57)
Years of education 14.8 (2.6) 14.9 (2.5) 14.9 (2.5)
n (%) n (%) n (%)
Sex
 Male 888 (44.3) 796 (44) 700 (44.1)
 Female 1115 (55.7) 1015 (56) 887 (55.9)
Race
 Black 848 (42.3) 781 (43.1) 656 (41.3)
 White 1155 (57.7) 1030 (56.9) 931 (58.7)
Adverse childhood experiences (ACEs) M (SD) M (SD) M (SD)
 Emotional neglect 1.81 (0.83) 1.81 (0.83) 1.80 (0.82)
 Emotional abuse 1.64 (0.87) 1.63 (0.87) 1.62 (0.86)
 Physical abuse 1.28 (0.64) 1.27 (0.63) 1.27 (0.63)
 Household substance use 1.62 (1.07) 1.61 (1.07) 1.61 (1.06)
 Household structure and supervision 1.75 (0.79) 1.73 (0.79) 1.74 (0.79)
Open in a new tab

Demographic and clinical characteristics of participants from the CARDIA cohort across the three analytic samples examining adverse childhood experiences (ACEs), proteomics, and cognition. ACE measures reflect frequency of childhood experiences; items were rated on a four-point Likert scale (higher values indicate greater adversity).

3.2. Associations between ACEs and circulating proteins

We identified 35 candidate proteins associated with at least one ACE subtype (p < 0.001, a priori screening threshold; Fig. 1, Supplemental Table 1). Emotional neglect and emotional abuse showed the fewest associations (3 and 5 proteins, respectively), including mostly negative associations with cell signaling adaptors (e.g., NENF, RSPO3) and metabolic regulators (e.g., adiponectin, LIPR2). Physical abuse was associated with 10 proteins, predominantly positive associations with epithelial integrity proteins (e.g., DLG4, KNBP1, keratin 18). Household substance use was associated with six multifunctional proteins involved in protein folding and proteostasis (MCTS1), immune processes (ABP1), enzymatic regulation (retinal dehydrogenase 1), and transcriptional regulation (TCEA3). Effect sizes (β) ranged from −0.039 to 0.072, mostly positive. Household structure and supervision exhibited the broadest proteomic signature (13 proteins). These proteins spanned all functional categories, including immune.

Fig. 1.

Fig. 1

Open in a new tab

Derivation of analytic samples from the CARDIA cohort. Flow diagram showing the derivation of the three analytic samples used in Analyses 1–3, beginning with the merged demographic and Year 25 proteomic dataset (N = 2231) and applying sequential exclusions for invalid demographic data and analysis-specific missingness. Blue boxes indicate intermediate dataset stages; red boxes indicate participants excluded at each step. Exclusion counts (n) are shown. Green boxes indicate the three final analytic samples used in Analyses 1–3. Analysis 4 (mediation analysis, gold box) tested all 35 candidate proteins identified in Analysis 1 across each ACE–protein–cognition pathway, within the same sample as Analysis 3 (ACEs–Cognition; N = 1587). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.3. Associations between ACE-related circulating proteins and cognition

Among the 35 ACE-related proteins, two were associated with cognition after FDR correction (q < 0.05) (Fig. 2, Supplemental Table 3). Higher MCTS1, a regulator of translation re-initiation and cell cycle progression, was associated with worse processing speed (β = −0.088, q = 0.002). Higher AINX, a neuron-specific intermediate filament protein, was associated with worse executive function (β = −0.069, q = 0.025).

Fig. 2.

Fig. 2

Open in a new tab

Plasma proteomic signatures associated with adverse childhood experiences (ACEs). Volcano plots showing adjusted linear regression results of effect sizes (β) for the association between individual ACE types and plasma protein levels in the CARDIA cohort. Each point represents a single protein. Proteins with p < 0.05 are in orange; proteins with p < 0.001 are in red and labeled by gene name. Gray points indicate proteins that do not meet statistical significance. Positive β values indicate higher protein levels associated with the ACE, and negative β values indicate lower protein levels. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.4. Associations between ACEs and cognition

Emotional neglect, household substance use, and household structure and supervision were associated with lower verbal memory (β = −0.089, β = −0.083, and β = −0.081, respectively), as was emotional abuse (β = −0.083) (all q < 0.05; Fig. 3 Supplemental Table 5). Household structure and supervision was also associated with lower executive function (β = −0.073, q = 0.007). No ACE had a significant association with processing speed or global cognition after correction (see Fig. 4).

Fig. 3.

Fig. 3

Open in a new tab

Correlation coefficients and adjusted regression beta effects between ACEs and cognitive performance. (A) Heatmap of raw Pearson correlation coefficients (r) between each ACE domain and cognitive outcome. (B) Heatmap of standardized regression coefficients (β) from linear models adjusted for age, sex, race, and years of education, with Benjamini-Hochberg false discovery rate (FDR) correction applied within the 20-test family. Both panels reflect the same analytic sample (N = 1587), restricted to participants with complete data on ACEs, cognitive outcomes, and all covariates. Cell values show the coefficient and significance level; color indicates direction and magnitude, with blue representing negative associations and red representing positive associations. Panel A significance reflects uncorrected two-sided p-values; Panel B significance reflects FDR-corrected q-value. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4.

Fig. 4

Open in a new tab

Associations between ACEs and cognitive performance. Forest plot of standardized regression coefficients (β) and 95% confidence intervals for associations between each ACE domain and cognitive outcome (20 tests total: 5 ACE domains × 4 cognitive domains), adjusted for age, sex, race, and years of education. Filled points indicate associations surviving Benjamini-Hochberg false discovery rate (FDR) correction (q < 0.05); open points did not survive correction. The dashed vertical line at zero indicates no association. Point colors indicate ACE domain. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.5. Mediation of ACE-cognition associations by candidate proteins

In FDR-corrected mediation analyses testing all 35 candidate proteins across each ACE-cognition pathway (700 tests total), no pathway reached statistical significance after correction (q < 0.05; Supplemental Table 7). Among the strongest signals, not FDR-significant, were household substance use via MCTSI on processing speed (ACME = 0.106, proportion mediated = −49.3%, p = 0.0002), and physical abuse via AINX on executive function (ACME = 0.141, proportion mediated = −13% [suppression pattern rather than conventional mediation], p = 0.0004). These same two proteins were independently associated with cognition in the FDR-corrected protein-cognition analysis above.

3.6. Stratification associations by age, sex, and race

Given the exploratory nature of these subgroup analyses, we also applied FDR correction within each stratified test family (Supplemental Tables 2, 4, 6). 84 of 1050 ACE-protein associations, 14 of 120 ACE-cognition associations, and 128 of 840 protein-cognition associations were significant (all q < 0.05). By age, older midlife participants showed a positive association between household structure and supervision and BICD1, a cytoskeletal adaptor involved in intracellular transport (β = 0.098, q = 0.002), and a positive association between adiponectin and verbal memory (β = 0.262, q < 0.001). By sex, female participants showed the strongest association between household substance use and retinal dehydrogenase 1 (β = 0.088, q = 0.001), MCTS1 with processing speed (β = −0.2, q < 0.001), and adiponectin with global cognition (β = 0.233, q < 0.001). By race, Black participants showed the largest-magnitude association between physical abuse and executive function (β = −0.105, q = 0.044), as well as associations between physical abuse and both DLG4 (β = 0.150, q = 0.002) and AKIR2, an immune signaling regulator (β = 0.095, q = 0.006). Raw distributions of all ACE and cognitive domains, stratified by age, race, sex, and education, are shown in Fig. 5.

Fig. 5.

Fig. 5

Open in a new tab

Distribution of ACE and cognitive scores by age, race, sex, and years of education. Density plots showing raw score distributions by demographic subgroup (age: <55 vs. ≥55; race: Black vs. White; sex: female vs. male; education: ≤12 vs. >12 years), distinguished by color and line style. (A) The five ACE domains, on a shared 1–4 scale. (B) The four cognitive domains, each on its own scale. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

4. Discussion

This study provides the first comprehensive proteomic characterization of biological embedding associated with ACEs in a biracial, community-based cohort. Childhood adversity was associated with distinct midlife proteomic signatures and cognition, with notable differences across demographic groups. Two proteins, MCTS1, a regulator of translation re-initiation and cell cycle progression, and AINX, a neuron-specific intermediate filament protein, were both identified among the candidate proteins associated with ACEs in our initial screen, and both were associated with cognition after correcting for multiple comparisons. Although neither protein was a significant mediator of the ACE-cognition association, this convergence across independent analytic steps makes them exploratory candidates warranting independent replication. Identifying midlife proteomic signatures may clarify early biological changes that precede clinical disease, support risk stratification, and inform preventive strategies for populations exposed to ACEs.

Of the 35 candidate proteins associated with ACEs, most spanned categories unrelated to immune signaling. Metabolic and enzymatic regulation accounted for the largest share, including adiponectin, several lipid-metabolism enzymes, and retinal dehydrogenase 1. Cell signaling adaptors and stress-responsive modulators made up the next largest group, including AINX. The remaining candidates spanned gene expression and RNA processing, epithelial integrity, and protein folding and proteostasis. Only three proteins, ABP1, AKIR2, and PSME2, fell into a formally immune-related category. Prior research has focused on canonical (e.g., CRP) (Danese et al., 2007) and other inflammatory markers (Bourassa et al., 2021; Coelho et al., 2014), and our stratified analyses found that ACEs were associated with immune-related proteins in specific demographic subgroups, though not in the pooled sample. Physical abuse was associated with higher AKIR2 (a protein involved in immune adhesion and vascular inflammatory signaling) among older midlife, Black, and female participants, and AKIR2 in turn was associated with better global cognition in Whites and worse executive function in female participants. PSME2, a subunit of the immunoproteasome involved in MHC class I antigen processing and gamma-interferon signaling, showed a similarly robust stratified signature: household substance use was associated with higher PSME2 among younger midlife, Black, and female participants, and higher PSME2 was associated with worse processing speed across younger midlife, older midlife, Black, and White strata.

In the pooled, non-stratified sample, however, only two proteins were associated with cognition after correction for multiple comparisons: higher MCTS1, a regulator of translation re-initiation and cell cycle progression, was associated with worse processing speed, and higher AINX, a neuron-specific intermediate filament protein, was associated with worse executive function. Neither is an immune signaling protein, which suggests that translational control and neuronal cytoskeletal integrity may be more robust proteomic correlates of ACE-related cognitive differences in the full cohort. Immune-related signaling may be more prominent within specific demographic subgroups (Baram and Birnie, 2024).

Among associations between ACEs and cognition, verbal memory was the most consistently impaired domain (four out of 5 ACEs). This fits with prior evidence linking early-life adversity to differences in stress-sensitive memory systems (Baram and Birnie, 2024). Household structure and supervision was additionally associated with lower executive function. Contrary to the initial hypothesis, no ACE-protein-cognition pathway was significant in mediation analyses. However, the strongest candidate pathways, household substance use via MCTS1 on processing speed and physical abuse via AINX on executive function, are worth noting. Although they did not mediate the ACE-cognition association in the present study, MCTS1 and AINX may still be the most promising candidate proteins for future mediation work (Bohlen et al., 2023; Prosniak et al., 1998).Prior studies show that MCTS1 influences cell cycle regulation and translational efficiency (Prosniak et al., 1998; Shi et al., 2003). Its activation under chronic stress may represent a complex balance between adaptive compensation and longer-term biological cost. Together with AINX's role in neuronal cytoskeletal integrity, these mechanisms offer plausible biological rationale for both candidates, though replication in larger, independent cohorts with adequately powered mediation designs will be needed to determine if either protein plays a causal role in tying childhood adversity to cognition.

In sex-stratified analyses, the most frequent ACE–protein and protein–cognition associations in both females and males involved proteins of metabolic and enzymatic regulation and cell signaling adaptation. Among females, household substance use in particular was associated with increased retinal dehydrogenase 1, an enzyme that regulates retinoic acid synthesis and influences gene transcription, synaptic plasticity, and neurodevelopmental signaling (Piazza et al., 2024). This suggests that ACEs in females may preferentially engage retinoid-dependent transcriptional programs involved in synaptic remodeling and cognitive adaptation, with retinal dehydrogenase 1 as a peripheral signature of this broader neurobiological process (Piazza et al., 2024). Both sexes shared metabolic and cell signaling protein alterations in response to ACEs, but females showed preferential engagement of this synaptic plasticity-related metabolic protein.

In race-stratified analyses, Black participants exhibited a pronounced association of physical abuse with AKIR2, a protein involved in immune signaling regulation, as well as DLG4, a postsynaptic scaffolding protein important for synaptic organization. 4950Physical abuse also showed its strongest and most negative association with executive function in Black participants. Physical abuse can impair executive function by disrupting prefrontal cortical development, altering stress-responsive neural circuits, and dysregulating neurotransmitter systems critical for cognitive control (De Bellis and Zisk, 2014; De Bellis et al., 1999). These effects manifest across executive function domains, including set-shifting, cognitive control, and non-verbal reasoning (Assuras et al., 2025; Maxfield et al., 2026). These deficits persist into middle and older adulthood even after accounting for psychiatric comorbidities (Maxfield et al., 2026). The associations of physical abuse with both higher AKIR2 and worse executive function suggests that altered immune signaling is one route through which social adversity may affect brain function. These findings are exploratory, within-group associations. Because race interactions and race-stratified mediation analyses were not formally conducted, AKIR2 should not be interpreted as representing a race-specific biological route from physical abuse to executive function. In this cohort, race was self-identified. These observed patterns should not be interpreted as reflecting intrinsic biological differences between racial groups.

Several limitations to our study warrant consideration. First, retrospective assessment of ACEs introduces potential recall bias, and the CARDIA questionnaire may not have captured the full complexity of ACEs. Second, our sample included only Black and White participants from select US regions, which limits generalizability. Third, because global cognition (MoCA) was assessed only at Year 30, Year 25 cognitive performance could not be adjusted for, and thus cannot distinguish ACE-related differences in attained cognitive level at Year 30 from ACE-related change over the Year 25-Year 30 interval. This distinction is a priority for follow-up analysis in this cohort. Fourth, the SomaScan platform comprehensively measures circulating protein levels that may not fully capture tissue-specific protein expression or activity. One additional consideration is the lack of adjustment for midlife conditions such as depression or cardiovascular disease. However, these are more likely to mediate the causal pathway from ACEs to biological or cognitive outcomes, and adjusting for them raises concerns about obscuring total ACE effects. Finally, the immune signal in our mediation analyses was anchored primarily by AKIR2, though proteins with immune-adjacent functions, including PSME2 and CUL3, appeared at earlier steps of the ACE–protein–cognition pathway. Future work with larger samples and immune-focused panels could better characterize the full scope of immune involvement in these pathways.

Our findings open several important avenues for future research. First, longitudinal studies tracking ACE-related proteomic signatures from early adulthood through older age could identify critical periods for intervention of ACE-related sequelae (e.g., psychosocial distress) and clarify how biological embedding evolves over time. Second, mechanistic studies examining the relationships between proteins and cognitive processes could help establish causal pathways and therapeutic targets. Studies that test the mediation of immune protein signatures, particularly AKIR2, in the association between social adversity and social-cognitive outcomes would speak directly to a social psychoneuroimmunology framework. Third, expanding these analyses to other omics platforms (e.g., genomics, metabolomics, transcriptomics) could provide a broader view of biological embedding. Finally, studies examining if similar age-related patterns appear in other ACE-associated health outcomes could determine if these findings extend beyond cognition.

This work has significant implications for precision medicine approaches to prevent and treat the long-term cognitive consequences of childhood adversity. Identifying candidate proteins may provide better characterizations of the biological pathways that associate childhood adversity with cognition. Demonstrating that biological embedding can be detected and quantified through proteomic profiling raises the possibility that, with further validation, protein signatures could contribute to future risk stratification efforts. More broadly, these findings position altered protein signaling as one component of a broader biological signature pathways connecting early-life adversity to the social-cognitive capacities that support everyday functioning in adulthood.

CRediT authorship contribution statement

Deborah K. Rose: Conceptualization, Formal analysis, Methodology, Software, Visualization, Writing – original draft, Writing – review & editing. Gabrielle N. Pfund: Conceptualization, Formal analysis, Methodology, Software, Visualization, Writing – review & editing. Robin Ortiz: Conceptualization, Supervision, Writing – review & editing. Sithara Vivek: Methodology, Writing – review & editing. Roland J. Thorpe Jr: Writing – review & editing. Kristine Yaffe: Writing – review & editing. Indira C. Turney: Writing – review & editing. Keenan A. Walker: Conceptualization, Methodology, Supervision, Writing – review & editing. Sarah N. Forrester: Conceptualization, Methodology, Supervision, Writing – original draft, Writing – review & editing.

Consent to participate/publication

Participant consent was obtained per CARDIA protocols.

Data availability

Due to participant privacy, individual-level omics and cognitive data cannot be publicly shared; summary data may be available upon request.

Funding

This research was partly supported by the Intramural Research Program of the National Institutes of Health (NIH). The contributions of the NIH author(s) are considered works of the United States Government. The findings and conclusions presented in this paper are those of the author(s) and do not necessarily reflect the views of the NIH or the US Department of Health and Human Services.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

The Coronary Artery Risk Development in Young Adults Study (CARDIA) is supported by contracts 75N92023D00002, 75N92023D00003, 75N92023D00004, and 75N92023D00005, and 75N92023D00006. Collection of Year-30 cognitive data was funded by NHLBI grant R01HL122658. Dr. Thorpe was supported by NIH grant P30AG059298.

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.bbih.2026.101353.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Multimedia component 1
mmc1.docx (438.6KB, docx)

References

  1. Assuras S., Courtney K., Maxfield M., et al. Childhood maltreatment confers long-term risk for cognitive impairment: a prospective investigation. J Prev Alzheimers Dis. 2025;12(9) doi: 10.1016/J.TJPAD.2025.100303. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Baram T.Z., Birnie M.T. Enduring memory consequences of early-life stress/adversity: structural, synaptic, molecular and epigenetic mechanisms. Neurobiol. Stress. 2024;33 doi: 10.1016/J.YNSTR.2024.100669. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Baumeister D., Akhtar R., Ciufolini S., Pariante C.M., Mondelli V. Childhood trauma and adulthood inflammation: a meta-analysis of peripheral C-reactive protein, interleukin-6 and tumour necrosis factor-α. Mol. Psychiatr. 2016;21(5):642. doi: 10.1038/MP.2015.67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Beste C., Stock A.K., Zink N., Ocklenburg S., Akgün K., Ziemssen T. How minimal variations in neuronal cytoskeletal integrity modulate cognitive control. Neuroimage. 2019;185:129–139. doi: 10.1016/j.neuroimage.2018.10.053. [DOI] [PubMed] [Google Scholar]
  5. Bohlen J., Zhou Q., Philippot Q., et al. Human MCTS1-dependent translation of JAK2 is essential for IFN-γ immunity to mycobacteria. Cell. 2023;186(23):5114. doi: 10.1016/J.CELL.2023.09.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bourassa K.J., Rasmussen L.J.H., Danese A., et al. Linking stressful life events and chronic inflammation using suPAR (soluble urokinase plasminogen activator receptor) Brain Behav. Immun. 2021;97:79. doi: 10.1016/J.BBI.2021.06.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Candia J., Daya G.N., Tanaka T., Ferrucci L., Walker K.A. Assessment of variability in the plasma 7k SomaScan proteomics assay. Sci. Rep. 2022;12(1) doi: 10.1038/S41598-022-22116-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chen M.A., LeRoy A.S., Majd M., et al. Immune and epigenetic pathways linking childhood adversity and health across the lifespan. Front. Psychol. 2021;12 doi: 10.3389/FPSYG.2021.788351. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chiang J.J., Lam P.H., Chen E., Miller G.E. Psychological stress during childhood and adolescence and its association with inflammation across the lifespan: a critical review and meta-analysis. Psychol. Bull. 2022;148(1–2):27. doi: 10.1037/BUL0000351. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Choi B., Liu G.Y., Sheng Q., et al. Proteomic biomarkers of quantitative interstitial abnormalities in COPDGene and CARDIA lung study. Am. J. Respir. Crit. Care Med. 2024;209(9):1091–1100. doi: 10.1164/rccm.202307-1129OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Coelho R., Viola T.W., Walss-Bass C., Brietzke E., Grassi-Oliveira R. Childhood maltreatment and inflammatory markers: a systematic review. Acta Psychiatr. Scand. 2014;129(3):180–192. doi: 10.1111/ACPS.12217. [DOI] [PubMed] [Google Scholar]
  12. Danese A., McEwen B.S. Adverse childhood experiences, allostasis, allostatic load, and age-related disease. Physiol. Behav. 2011;106(1) doi: 10.1016/j.physbeh.2011.08.019. [DOI] [PubMed] [Google Scholar]
  13. Danese A., Moffitt T.E., Harrington H.L., et al. Adverse childhood experiences and adult risk factors for age-related disease: depression, inflammation, and clustering of metabolic risk markers. Arch. Pediatr. Adolesc. Med. 2009;163(12):1135. doi: 10.1001/ARCHPEDIATRICS.2009.214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Danese A., Pariante C.M., Caspi A., Taylor A., Poulton R. Childhood maltreatment predicts adult inflammation in a life-course study. Proc. Natl. Acad. Sci. U. S. A. 2007;104(4):1319–1324. doi: 10.1073/PNAS.0610362104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Danese A, Widom CS. Objective and Subjective Experiences of Child Maltreatment and their Relationships with Psychopathology. doi:10.1038/s41562-020-0880-3. [DOI] [PubMed]
  16. Daskalakis N.P., Yehuda R. In: Programming the HPA-Axis by Early Life Experience: Mechanisms of Stress Susceptibility and Adaptation. Daskalakis N.P., Yehuda R., editors. Frontiers in Endocrinology; 2015. [Google Scholar]
  17. De Bellis M.D., Keshavan M.S., Clark D.B., et al. Developmental traumatology part II: brain development. Biol. Psychiatry. 1999;45(10):1271–1284. doi: 10.1016/S0006-3223(99)00045-1. [DOI] [PubMed] [Google Scholar]
  18. De Bellis M.D., Zisk A. The biological effects of childhood trauma. Child Adolesc Psychiatr Clin N Am. W.B. Saunders. 2014;23(2):185–222. doi: 10.1016/j.chc.2014.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Felitti V.J., Anda R.F., Nordenberg D., et al. Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: the adverse childhood experiences (ACE) study. Am. J. Prev. Med. 1998;14(4):245–258. doi: 10.1016/S0749-3797(98)00017-8. [DOI] [PubMed] [Google Scholar]
  20. Freitas S., Simões M.R., Alves L., Santana I. Montreal cognitive assessment: validation study for mild cognitive impairment and alzheimer disease. Alzheimer Dis. Assoc. Disord. 2013;27(1):37–43. doi: 10.1097/WAD.0b013e3182420bfe. [DOI] [PubMed] [Google Scholar]
  21. Friedman G.D., Cutter G.R., Donahue R.P., et al. CARDIA: study design, recruitment, and some characteristics of the examined subjects. J. Clin. Epidemiol. 1988;41(11):1105–1116. doi: 10.1016/0895-4356(88)90080-7. [DOI] [PubMed] [Google Scholar]
  22. Hetz C. Adapting the proteostasis capacity to sustain brain healthspan. Cell. 2021;184(6):1545–1560. doi: 10.1016/j.cell.2021.02.007. [DOI] [PubMed] [Google Scholar]
  23. Koprulu M., Wheeler E., Kerrison N.D., et al. Sex differences in the genetic regulation of the human plasma proteome. Nat. Commun. 2025;16(1):4001. doi: 10.1038/S41467-025-59034-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lam N., Fairweather S., Lewer D., et al. The association between adverse childhood experiences and mental health, behaviour, and educational performance in adolescence: a systematic scoping review. Cardona J.F., editor. PLOS Mental Health. 2024;1(5) doi: 10.1371/JOURNAL.PMEN.0000165. [DOI] [PubMed] [Google Scholar]
  25. Lehallier B., Gate D., Schaum N., et al. Undulating changes in human plasma proteome profiles across the lifespan. Nat. Med. 2019;25(12):1843. doi: 10.1038/S41591-019-0673-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lezak M.D., Hoieson D.B., Bigler E.D., Tranel D. fifth ed. Oxford University Press; 2012. Neuropsychological Assessment, 5th Ed.https://psycnet.apa.org/record/2012-02043-000 [Google Scholar]
  27. Lin J.E., Neylan T.C., Epel E., O'Donovan A. Associations of childhood adversity and adulthood trauma with C-reactive protein: a cross-sectional population-based study. Brain Behav. Immun. 2015;53:105. doi: 10.1016/J.BBI.2015.11.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Loucks E.B., Almeida N.D., Taylor S.E., Matthews K.A. Childhood family psychosocial environment and coronary heart disease risk. Psychosom. Med. 2011;73(7):563. doi: 10.1097/PSY.0B013E318228C820. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Loucks E.B., Taylor S.E., Polak J.F., Wilhelm A., Kalra P., Matthews K.A. Childhood family psychosocial environment and carotid intima media thickness: the CARDIA study. Soc. Sci. Med. 2013;104:15. doi: 10.1016/J.SOCSCIMED.2013.12.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Manly J.J., Jacobs D.M., Sano M., et al. Cognitive test performance among nondemented elderly African Americans and whites. Neurology. 1998;50(5):1238–1245. doi: 10.1212/WNL.50.5.1238. [DOI] [PubMed] [Google Scholar]
  31. Maxfield M., Courtney K., Assuras S., Manly J.J., Widom C.S. Childhood maltreatment and cognitive functioning from young adulthood to late midlife: a prospective study. Neuropsychology. 2026;40(2):178–190. doi: 10.1037/neu0001042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. McCarrey A.C., An Y., Kitner-Triolo M.H., Ferrucci L., Resnick S.M. Sex differences in cognitive trajectories in clinically normal older adults. Psychol. Aging. 2016;31(2):166. doi: 10.1037/PAG0000070. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Mersky J.P., Choi C., Plummer Lee C.T., Janczewski C.E. Disparities in adverse childhood experiences by race/ethnicity, gender, and economic status: intersectional analysis of a nationally representative sample. Child Abuse Negl. 2021;117 doi: 10.1016/J.CHIABU.2021.105066. [DOI] [PubMed] [Google Scholar]
  34. Muscatell K.A. Social psychoneuroimmunology: understanding bidirectional links between social experiences and the immune system. Brain Behav. Immun. 2021;93:1–3. doi: 10.1016/J.BBI.2020.12.023. [DOI] [PubMed] [Google Scholar]
  35. Na D., Zhang Z., Meng M., et al. Energy metabolism and brain aging: strategies to delay neuronal degeneration. Cell. Mol. Neurobiol. 2025;45(1) doi: 10.1007/S10571-025-01555-Z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Nasreddine Z.S., Phillips N.A., Bédirian V., et al. The Montreal cognitive assessment, MoCA: a brief screening tool for mild cognitive impairment. J. Am. Geriatr. Soc. 2005;53:695–699. doi: 10.1111/j.1532-5415.2005.53221.x. www.mocatest [DOI] [PubMed] [Google Scholar]
  37. Nelson C.A., Scott R.D., Bhutta Z.A., Harris N.B., Danese A., Samara M. Adversity in childhood is linked to mental and physical health throughout life. Br. Med. J. 2020;371 doi: 10.1136/bmj.m3048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Nusslock R., Miller G.E. Early-life adversity and physical and emotional health across the lifespan: a neuro-immune network hypothesis. Biol. Psychiatry. 2015;80(1):23. doi: 10.1016/J.BIOPSYCH.2015.05.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Ortiz R., Kershaw K.N., Zhao S., et al. Evidence for the association between adverse childhood family environment, child abuse, and caregiver warmth and cardiovascular health across the lifespan: the coronary artery risk development in young adults (CARDIA) study. Circ Cardiovasc Qual Outcomes. 2024;17(2) doi: 10.1161/CIRCOUTCOMES.122.009794. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Pan J., Zaff J.F., Porche M. Social support, childhood adversities, and academic outcomes: a latent class analysis. J. Educ. Stud. Placed A. T. Risk. 2019;25(3):251–271. doi: 10.1080/10824669.2019.1708744. [DOI] [Google Scholar]
  41. Piazza A., Carlone R., Spencer G.E. Non-canonical retinoid signaling in neural development, regeneration and synaptic function. Front. Mol. Neurosci. 2024;17 doi: 10.3389/FNMOL.2024.1371135/FULL. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Pierce J.B., Kershaw K.N., Kiefe C.I., et al. Association of childhood psychosocial environment with 30‐Year cardiovascular disease incidence and mortality in middle age. J. Am. Heart Assoc.: Cardiovascular and Cerebrovascular Disease. 2020;9(9) doi: 10.1161/JAHA.119.015326. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Prosniak M., Dierov J., Okami K., et al. A novel candidate oncogene, MCT-1, is involved in cell cycle progression. Cancer Res. 1998;58(19):4233–4237. [PubMed] [Google Scholar]
  44. Rosa M., Scassellati C., Cattaneo A. Association of childhood trauma with cognitive domains in adult patients with mental disorders and in non-clinical populations: a systematic review. Front Psychol. Frontiers Media SA. 2023;14 doi: 10.3389/fpsyg.2023.1156415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Rosa M., Scassellati C., Cattaneo A. Association of childhood trauma with cognitive domains in adult patients with mental disorders and in non-clinical populations: a systematic review. Front. Psychol. 2023;14 doi: 10.3389/FPSYG.2023.1156415/FULL. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Schmidt M. Western Psychological Services; 1996. Rey Auditory Verbal Learning Test: a Handbook.https://www.google.com/books/edition/Rey_Auditory_Verbal_Learning_Test/UOcPRAAACAAJ?hl=en [Google Scholar]
  47. Sharp E.S., Gatz M. The relationship between education and dementia an updated systematic review. Alzheimer Dis. Assoc. Disord. 2011;25(4):289. doi: 10.1097/WAD.0B013E318211C83C. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Shi B., Levenson V., Gartenhaus R.B. Identification and characterization of a novel enhancer for the human MCT-1 oncogene promoter. J. Cell. Biochem. 2003;90(1):68–79. doi: 10.1002/JCB.10609. [DOI] [PubMed] [Google Scholar]
  49. Tingley D., Yamamoto T., Hirose K., Keele L., Imai K. Mediation: r package for causal mediation analysis. J. Stat. Software. 2014;59(5):1–38. doi: 10.18637/JSS.V059.I05. [DOI] [Google Scholar]
  50. Wechsler D. Psychological Corp; 1955. Manual for the Wechsler Adult Intelligence Scale.https://psycnet.apa.org/record/1955-07334-000 [Google Scholar]
  51. Wechsler D. Harcourt Brace Jovanovich; 1987. WMS-R : Wechsler Memory Scale--Revised : Manual. Psychological Corp. [Google Scholar]
  52. Yan L.L., Liu K., Daviglus M.L., et al. Education, 15-year risk factor progression, and coronary artery calcium in young adulthood and early middle age: the coronary artery risk development in young adults study. JAMA. 2006;295(15):1793–1800. doi: 10.1001/JAMA.295.15.1793. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Multimedia component 1
mmc1.docx (438.6KB, docx)

Data Availability Statement

Due to participant privacy, individual-level omics and cognitive data cannot be publicly shared; summary data may be available upon request.

Articles from Brain, Behavior, & Immunity - Health are provided here courtesy of Elsevier

RESOURCES