Childhood maltreatment and cognitive functioning: the role of depression, parental education, and polygenic predisposition

The study addresses whether childhood maltreatment (CM) is associated with broad cognitive dysfunction independently of key confounders, including major depressive disorder (MDD), familial socioeconomic status (proxied by parental education, PE), and genetic predisposition for depression and educational attainment (EdA). Prior research links CM to increased risk and severity of psychiatric disorders, particularly MDD, and to cognitive impairments in working memory, attention, and intelligence, with mixed findings in episodic memory, processing speed, and executive function. Because MDD, SES, and genetic factors are each associated with cognitive performance and with CM exposure, the authors aim to disentangle their independent and joint effects. Hypotheses: CM and MDD (diagnosis and polygenic risk) negatively associate with cognition; PE and genetic predisposition for EdA positively associate with cognition; CM shows an independent effect on cognition after controlling for MDD, PE, and polygenic scores; exploratory tests probe interactions of CM with age, sex, MDD, PE, and genetic predisposition.

The introduction reviews meta-analytic and empirical evidence on CM-related cognitive deficits. Some meta-analyses report widespread, domain-nonspecific deficits among individuals with CM, while others find domain-specific effects (e.g., larger deficits in attention and visual episodic memory; inconsistent findings for executive function and visuospatial working memory). Cognitive deficits related to CM can be observed from childhood and persist into adulthood, and they appear in PTSD and non-PTSD samples. MDD shows overlapping cognitive deficits and higher prevalence among those with CM, complicating causal inference if not controlled. Familial SES is associated with both likelihood of CM and cognitive outcomes; earlier work often did not control for SES, though more recent large cohort studies indicated CM–cognition effects attenuate considerably when SES and early IQ are controlled, suggesting potential confounding. Genetic influences may underlie associations among SES, MDD liability, CM exposure, and cognition, as adoption and family studies indicate shared genetic influences. Gene–environment interplay has been proposed whereby familial liability for affective disorders elevates CM risk and its impact, warranting examination of polygenic scores for MDD and EdA in relation to cognition and CM.

Design and sample: Cross-sectional analysis from the FOR2107 Marburg–Münster Affective Disorders Cohort Study (MACS). Adults aged 18–65 with complete CM, diagnosis, PE, genetics, and at least one neurocognitive measure were included. Final N=1217: n=547 with MDD (current or lifetime DSM-IV major depressive episode) and n=670 healthy controls (HC). Mean age 34.7 years (SD 13.2); 62.64% female. All participants provided informed consent; exclusion criteria and details in Supplementary. Measures: - Childhood maltreatment: Childhood Trauma Questionnaire (CTQ) total and five subscales (emotional/physical/sexual abuse; emotional/physical neglect). - Socioeconomic background: Parental education (PE) via self-report, used as a proxy for familial SES. - Neurocognitive battery assessing: • Working memory: e.g., Corsi block tapping (forward/backward), Letter-Number Sequencing Test (LNST). • Executive functioning: Trail Making Test Part B (and A–B difference where applicable). • Processing speed: Trail Making Test Part A; Digit Symbol Substitution Test (DSST). • Sustained attention: measures related to TMT A (and attention-focused tasks, per Supplementary). • Declarative long-term/short-term memory: Verbal Learning and Memory Test (VLT/VLMT; immediate/short-term and delayed/long-term recall). • Verbal intelligence: Multiple-Choice Vocabulary Test (MWT-B). Test order and operationalization are detailed in Supplementary. - Clinical assessment: Structured Clinical Interview for DSM-IV (SCID-I) to ascertain diagnoses; Hamilton Depression Rating Scale (HDRS) for current depression severity. - Genetics: Genome-wide genotyping with standard QC; polygenic scores (PGS) computed for MDD and educational attainment (EdA) using LD clumping in PLINK v1.9 and GWAS summary statistics (details per prior consortium methods). PGS standardized. Statistical analyses: - Multivariate analyses across cognitive domains tested associations with CTQ sum, MDD diagnosis, PE, MDD PGS, and EdA PGS, controlling for age and sex. A base model and a full model including all predictors were used. Exploratory interactions of CTQ with age, sex, MDD, PE, and PGS were examined. - Univariate regressions for each cognitive domain assessed robustness when sequentially adding covariates (MDD diagnosis, PE, MDD PGS, EdA PGS). Multiple testing corrections (e.g., Bonferroni) were applied. Effect sizes reported as partial eta squared (η²_p), F-tests, and standardized betas; additional details in Supplementary. Sample characteristics (selected): Compared to HC, MDD participants showed higher CTQ total and subscale scores and worse cognitive performance across most tasks; detailed means/SDs and group comparisons are listed in Table 1 (e.g., CTQ total markedly higher in MDD; DSST lower in MDD; etc.).

- Multivariate effects across cognitive domains (base analyses): Significant associations with CM (η²≈0.083, P<0.001), PE (η²≈0.085, P<0.001), MDD PGS (η²≈0.021, P=0.005), and EdA PGS (η²≈0.031, P<0.001). These associations remained significant in the full model including all predictors. - Full multivariate model (including all variables) showed: • CTQ sum: F(10,1183)=2.763, P=0.002, η²_p=0.023 (worse cognition with higher CM). • MDD diagnosis: F(10,1183)=5.537, P<0.001, η²_p=0.045 (MDD associated with worse cognition). • Parental education: F(10,1183)=7.178, P<0.001, η²_p=0.057 (higher PE associated with better cognition). • MDD PGS: F(10,1183)=2.267, P=0.013, η²_p=0.018 (higher genetic risk linked to worse cognition, notably visuospatial working memory). • EdA PGS: F(10,1183)=2.139, P=0.019, η²_p=0.018 (higher genetic propensity linked to better cognition in several domains). • Interactions: Significant CTQ×sex (F(10,1183)=2.188, P=0.016, η²_p=0.018) and CTQ×age (F(10,1183)=1.977, P=0.033, η²_p=0.016). No significant CTQ×MDD interaction (all P>0.07). - Univariate domain findings: • CM associated with poorer performance across all cognitive domains in base models. • Adding MDD diagnosis rendered CM effects non-significant for TMT-A (processing speed/attention) and VLT-B (one memory measure) but CM remained significant for other domains (all P<0.024). The mean reduction in CTQ effect from base to model including MDD was ~36%, ranging from ~19% (MWT-B) to ~56% (TMT-B). • Adding PE reduced CM effects (mean beta reduction ~17% vs base); CM–VLT-B association became non-significant; PE showed positive main effects across all domains (all P<0.014). • Adding MDD PGS did not qualitatively alter CM effects; EdA PGS positively associated with working memory, attention, short-term memory, and verbal intelligence; MDD PGS negatively associated particularly with visuospatial working memory. Polygenic variables did not confound or moderate CM–cognition associations. - Overall pattern: CM, MDD, PE, and polygenic predispositions each independently contribute to cognitive variation; CM effects persist after adjustment but are attenuated, especially by MDD and PE.

The study demonstrates that CM is linked to broad cognitive deficits and that this relationship remains after stringent adjustment for MDD diagnosis, familial SES (PE), and polygenic predispositions for MDD and EdA. However, a substantial portion of the observed CM–cognition association is explained by MDD and PE, indicating both psychopathological and socioeconomic pathways to cognitive impairment. The lack of moderation by MDD or PE suggests the detrimental association of CM with cognitive performance is comparably evident across diagnostic status and across parental educational contexts. Polygenic influences for MDD and EdA show small but significant associations with cognition, yet they do not explain away nor moderate CM–cognition associations in this sample, suggesting environmental CM effects on cognition are not driven by these polygenic propensities. Findings support the notion that part of CM’s impact on cognition may be indirect via increased vulnerability to MDD, which itself is associated with cognitive dysfunction, while a direct CM effect likely remains. The observed sex- and age-related moderation indicates heterogeneity in CM-related cognitive outcomes, warranting targeted investigation. Clinically, even small-to-medium cognitive effects can have meaningful functional consequences in MDD, highlighting the need to assess and address cognition in individuals with CM histories.

This work clarifies that childhood maltreatment, depression, parental education, and polygenic predispositions for depression and educational attainment each contribute independently to cognitive functioning. CM is associated with widespread cognitive deficits that persist after controlling for these factors, albeit attenuated, especially by MDD and PE. The findings support comprehensive assessment of CM history, depressive psychopathology, SES background, and genetic risk when studying or treating cognitive dysfunction. Future research should: (1) use longitudinal designs to disentangle causal pathways and mediation by incident MDD; (2) refine SES measurement beyond parental education; (3) examine alternative genetic indices and candidate pathways, including potential gene–environment correlations; and (4) test tailored interventions (e.g., depression treatment combined with cognitive remediation/training) optimized for individuals with CM-related cognitive vulnerabilities.

- Cross-sectional design precludes causal inference; third-variable confounding remains possible. - CM measured via retrospective self-report (CTQ), which may be biased by mood/state or recall. - SES proxied by parental education only; other SES components (income, occupation, neighborhood) were not captured and may differentially relate to cognition. - Polygenic scores are limited to specific GWAS and methods; other genetic indices might yield different results. - Some measurement heterogeneity and potential task-specific sensitivities; detailed operationalizations relegated to Supplementary. - Generalizability limited to adult samples from the MACS cohort; inclusion of MDD cases and HC may not reflect broader community samples.