Gray matter correlates of childhood maltreatment lack replicability in a multi-cohort brain-wide association study

Abstract

Childhood maltreatment effects on cerebral gray matter have been frequently discussed as a neurobiological pathway for depression. However, localizations are highly heterogeneous, and recent reports have questioned the replicability of mental health neuroimaging findings. Here, we investigate the replicability of gray matter correlates of maltreatment, measured retrospectively via the Childhood Trauma Questionnaire, across three large adult cohorts (total N = 3225). Pooling cohorts yields maltreatment-related gray matter reductions, with most extensive effects when not controlling for depression diagnosis (maximum partial R² = .022). However, none of these effects significantly replicate across cohorts. Non-replicability is consistent across a variety of maltreatment subtypes and operationalizations, as well as subgroup analyses with and without depression, and stratified by sex. Results are furthermore consistent across a variety of gray matter operationalizations, including voxel-based morphometry and parcellation-based cortical and subcortical measures. In this work, we show that there is little evidence for the replicability of gray matter correlates of childhood maltreatment, when adequately controlling for psychopathology. This underscores the need to focus on replicability research in mental health neuroimaging.

Fig. 1. Associations between CTQ scales, demographic variables and clinical variables.

Note. a Spearman correlations are shown (two-sided). All correlations involving clinical variables (HDRS-17, BDI, Age of onset, number of depressive episodes and number of inpatient treatments) were only calculated within the MDD subsample. The BDI was only available within MACS and MNC, while the CES-D was only available for the BiDirect cohort. Significant associations at p < .05 are shown in bold font. b Violin plots are shown depicting the distribution of the CTQ sum scale, as well as the five CTQ subscales (based on total N = 3225; HC n = 1898; MDD n = 1327). Boxplots within the violins depict the median (center line), the interquartile range (IQR; box from the 25th to 75th percentile), and whiskers extending to the most extreme values within 1.5 × IQR. CTQ, childhood trauma questionnaire; EA, emotional abuse; PA, physical abuse; SA, sexual abuse; EN, emotional neglect; PN, physical neglect; HDRS-17, 17-item Hamilton Depression Rating Scale; BDI, Beck Depression Inventory; CES-D, Center for Epidemiologic Studies Depression Scale; HC, healthy control; MDD, major depressive disorder.

Fig. 2. Significant clusters from pooled analysis at p_FWE < .05.

Note. Glass brains are shown with maximum intensity projections. Color bars represent the partial R² of the CM predictor in the respective model (in Model 17 of the interaction term). Results were derived from general linear models with a one-sided test of the respective maltreatment predictor and a two-sided test of the interaction terms, both at p_FWE < .05. HC, healthy controls; MDD, major depressive disorder.

Fig. 3. Significant clusters across cohort-wise analyses at p_unc < .001 for exemplary models (replicability analysis).

Note. Glass brains are shown with maximum intensity projections. Color bars represent the partial R² of the maltreatment predictor in the respective model. Exemplary model results are shown for two models without any spatial overlap (i.e., replicability), namely the CTQ sum analysis in HC participants (Model 3) and in MDD participants (Model 4), as well as one model showing some degree of spatial overlap (Model 11 testing the physical neglect subscale as a predictor). Results were derived from general linear models with a one-sided test of the respective maltreatment predictor, at uncorrected p < .001. HC, healthy controls; MDD, Major Depressive Disorder; MACS, Marburg Münster Affective Disorders Cohort Study; MNC, Münster Neuroimaging Cohort; BiDirect, BiDirect cohort.