Common and distinct neural underpinnings of the association between childhood maltreatment and depression and aggressive behavior

Abstract

Background

Although childhood maltreatment (CM) is widely recognized as a transdiagnostic risk factor for various internalizing and externalizing psychological disorders, the neural basis underlying this association remain unclear. The potential reasons for the inconsistent findings may be attributed to the involvement of both common and specific neural pathways that mediate the influence of childhood maltreatment on the emergence of psychopathological conditions.

Methods

This study aimed to delineate both the common and distinct neural pathways linking childhood maltreatment to depression and aggression. First, we employed Network-Based Statistics (NBS) on resting-state functional magnetic resonance imaging (fMRI) data to identify functional connectivity (FC) patterns associated with depression and aggression. Mediation analyses were then conducted to assess the role of these FC patterns in the relationship between childhood maltreatment and each outcome.

Results

The results demonstrated that FC within the default mode network (DMN) and between the cingulo-opercular network (CON) and dorsal attention network (DAN) mediated the association between childhood maltreatment and aggression, whereas FC within the reward system and between the CON and the reward system mediated the link between childhood maltreatment and depression.

Conclusions

We speculate that the control system may serve as a transdiagnostic neural basis accounting for the sequela of childhood maltreatment, and the attention network and the reward network may act as specific neural basis linking childhood maltreatment to depression and aggression, respectively.

Introduction

Childhood maltreatment (CM) encompasses the adverse and traumatic experiences encountered by individuals during their formative years, including various forms of abuse and neglect, such as physical, sexual, and emotional abuse, as well as neglect [1, 2]. Childhood exposure to maltreatment is both highly prevalent and constitutes a significant global moral and public health issue [3, 4]. Approximately one billion children globally endure emotional abuse, physical abuse, sexual abuse, emotional neglect, and physical neglect annually [5]. A systematic review revealed that Chinese children experienced physical abuse, emotional abuse, sexual abuse, and emotional neglect at prevalence rates of 26.6%, 19.6%, 8.7%, and 26%, respectively [6]. Chil maltreatment has been demonstrated to be a significant risk factor for internalizing and externalizing problems across the lifespan, including depression [7], anxiety [8], aggression [9], and self-injury [10].

Given the robust association between childhood maltreatment and psychopathological issues, a substantial body of research has sought to investigate the neurobiological consequences of childhood maltreatment. The findings implicate multiple brain systems involved in threat processing (amygdala) [11, 12], reward processing (striatum) [13, 14], executive control (dorsolateral prefrontal cortex, DLPFC; dorsal anterior cingulate cortex, dACC) [15, 16] and autobiographical memory (hippocampus, posterior cingulate cortex, PCC) [17, 18]. In addition, studies suggested that abnormalities in these brain areas were not only the neural sequelae of childhood maltreatment, but also the potential risks responsible for the development psychopathological symptoms following exposure to early life adversity [19–21]. For instance, the early life stress response is characterized by the activation of the hypothalamic-pituitary-adrenal (HPA) axis. Hyperactivity of the HPA axis leads to increased cortisol levels, which subsequently impact brain regions involved in emotion processing and cognitive control, thereby exacerbating emotional disorders and behavioral problems [22, 23]. Bounoua and colleagues identified evidence suggesting that reduced cortical thickness in the prefrontal region, particularly around the left orbitofrontal cortex (OFC), may serve as an intermediary mechanism linking childhood trauma to an increased risk of future aggressive behavior [24]. A resting-state functional magnetic resonance imaging (fMRI) study indicated that increased exposure to stressful life events was inversely correlated with functional connectivity (FC) strength between the bilateral amygdala and the bilateral medial prefrontal cortices (mPFC), and these reduced FCs were significantly associated with elevated levels of aggressive behaviors and attentional difficulties [25]. Furthermore, childhood maltreatment could disrupt normal prefrontal interactions with the reward system, potentially leading to impulsive or addictive behaviors [19, 20, 26]. Hanson and others demonstrated that the smaller amygdala and hippocampal volumes were associated with greater cumulative stress exposure and behavior problems, hippocampal volumes partially mediated the relationship between early life stress and greater behavior problems, and this study suggested stress may shape the development of brain areas involved with emotion processing [27].

Meanwhile, an extensive body of neuroimaging research has elucidated the neural mechanism underlying the effects of childhood maltreatment on depression [28–30]. Key mechanisms implicated include the reward processing system, the threat processing system, the cingulo-opercular network (CON) and the default mode network (DMN) [27, 31–33]. Nagy’s fMRI study on facial emotion recognition revealed altered anterior striatal reward circuit function in major depressive disorder (MDD) and childhood maltreatment patients, with abused depression patients showing reduced right nucleus accumbens and cingulate cortex BOLD responses, and increased precentral and postcentral gyrus activity [34]. A longitudinal study by Rakesh et al. examined the role of increased connectivity, particularly among the DMN, the fronto-parietal network (FPN), the dorsal attention network (DAN), and the salience network (SAN), in mediating the relationship between childhood maltreatment and increased depression symptoms [32]. Mao et al. using cognitive reappraisal fMRI task posited that the maltreated group showed reduced OFC activation and enhanced FC between DLPFC, PCC, OFC, and amygdala during cognitive reappraisal relative to controls, and the increased FC between the DLPFC and the amygdala serves as a mediator in the childhood maltreatment-depression link [35]. A fMRI study elucidated distinct neural abnormalities in the MDD depending on prior childhood maltreatment history, particularly potentiated stress-related mPFC activation among MDD individuals reporting childhood maltreatment. Moreover, childhood maltreatment history was generally associated with the disruption in FC centered on the mPFC [36]. Furthermore, exposure to stressful life events was correlated with reduced hippocampal and amygdalar volumes, with the gray matter volume within pivotal nodes of the neural threat system (i.e., hippocampus and amygdala) identified as a mediating factor in the association between childhood trauma and the onset of depression [33].

Previous studies have emphasized that the potential mechanisms underlying the influence of childhood maltreatment on aggression and depression involve multiple systems, including the reward processing system, the threat processing system, and the executive control system. Nevertheless, these findings are confusing. Some studies have identified abnormalities in the threat system as mechanisms for internalizing problems associated with childhood maltreatment [37, 38]. However, the generalization of hyperactivity to threatening stimuli has also been proposed as a factor that may facilitate the emergence of aggressive behavior in individuals exposed to childhood trauma [27, 39]. Furthermore, an additional study has indicated that enhanced activity within the reward circuit is associated with an increased risk of depression following early adverse experiences. In contrast, other research has focused on the blunted neural response in the reward system as a mechanism for externalizing problems, such as aggression, that are linked to childhood maltreatment. In addition, several studies have examined the cognitive processing characteristics that are prevalent in both internalizing and externalizing problems [40–42].

Collectively, the findings from prior research are ambiguous. The investigations have not yielded consistent conclusions regarding the potential relationship linking childhood maltreatment to aggression and depression. This ambiguity may stem from two primary factors. One factor is that the relationship between childhood maltreatment and aggression as well as depression may involve both common and specific neural underpinnings. Another factor is that the limited sample sizes in these studies may restrict the generalizability of the findings. Therefore, building upon the findings of earlier investigations, we speculate that the DMN and the control system shared neural substrates in the association between childhood maltreatment and aggression/depression, the attentional network serves as a distinct neural correlate specifically linked to the relationship between childhood maltreatment and aggressive behaviors, while the reward system is a distinct neural substrate implicated in the correlation between childhood maltreatment and depression. The current study sought to elucidate the common and distinct neural bases through which childhood maltreatment links depression and aggressive behavior, utilizing a substantial sample of resting-state fMRI data. Initially, we employed Network-Based Statistics (NBS) to identify neuromarkers associated with aggression and depression, respectively. Subsequently, mediation models were implemented to assess whether these FC patterns could account for the relationships between childhood maltreatment and both aggression and depression. Finally, we testing our hypothesis conducted a comparative analysis of these FC patterns to delineate the common and distinct mechanisms linking childhood maltreatment to aggression and depression.

Methods

Participants

The data in this study were drawn from a large Gene-Brain-Behavior (GBB) project. The recruitment program and exclusion criteria are detailed in our previous publications [43, 44]. In the project, these participants finished fMRI scanning and were evaluated using behavioral assessments such as childhood maltreatment, aggression and depression. Participants with excessive head motion (mean fd power > 0.2) and those did not have behavioral data were excluded. Specifically, among the 342 undergraduate students who completed the Buss Perry Aggression Questionnaire (BPAQ), 254 students (mean age 19.94 ± 1.44 years; range: 17–26 years; 187 females and 67 males) finished the Child Trauma Questionnaire-Short Form (CTQ-SF) and imaging data, 103 of them in moderate to severe of childhood maltreatment (74.76% females and 25.24% males). 518 undergraduate students (mean age 19.32 ± 1.46 years; range: 16–26 years; 372 females and 146 males; moderate to severe of childhood maltreatment individuals: 226; moderate to severe of depression individuals: 87) finished the Beck Depression Inventory-II (BDI-II), Child Trauma Questionnaire-Short Form, and imaging data, 226 of them in moderate to severe of childhood maltreatment (69.33% females and 30.67% males). All participants had no psychopathological symptoms or neurological diseases and provided written informed consents prior experiments, and the experimental procedures approved by the ethical committee at the Southwest University in accordance with the Declaration of Helsinkiwere.

Behavior measures

Child Trauma Questionnaire-Short Form [45]. It is a widely used retrospective self-report instrument designed to assess various types of childhood adversity. This includes emotional, physical, and sexual abuse, as well as emotional and physical neglect. Participants were instructed to answer 28-items on Likert 5 point, ranging from “never” to “always” [45]. Based on the guidelines from Bernstein et al.’ s self-report questionnaire, individuals are classified as having experienced moderate to severe childhood trauma if they score ≥ 13 for emotional abuse, ≥ 10 for physical abuse, ≥ 8 for sexual abuse, ≥ 15 for emotional neglect, or ≥ 10 for physical neglect on any subscale [46]. The CTQ-SF has demonstrated strong reliability and validity in previous research, with a Cronbach’s alpha coefficient of 0.824, indicating excellent construct validity [45].

Buss-Perry Aggression Questionnaire [47]. Participants completed a self-reported measure of aggression comprising 29 items rated on a 1–5 scale, ranging from “extremely uncharacteristic of me” to “extremely characteristic of me” [47]. The BPAQ is composed of four subscales: physical aggression, verbal aggression, angry, and hostility. Previous research has demonstrated the internal consistency and test-retest reliability of the BPAQ. The Cronbach’s alpha for the total score was 0.93 [47].

Beck Depression Inventory-II [48]. It is a 21-item self-report questionnaire employed to evaluate the severity of depression symptoms experienced over the previous fortnight. Each items within the BDI-II was rated on a 4-point Likert scale ranging from 0 to 3. The cumulative score was then categorized as follows: 0–13 indicating no depression, 14–19 suggesting mild depression, 20–28 signaling moderate depression, and 29–63 denoting severe depression [48]. The Chinese version of the Beck Depression Inventory-II was utilized in this research. The Cronbach’s alpha for the BDI-II was 0.86, indicating a high level of internal consistency in measuring an underlying dimension of depression [49].

fMRI data acquisition and analysis

Image acquisition and preprocessing

Resting-state fMRI data were obtained using a Siemens 3T Trio scanner (Siemens Medical Systems, Erlangen, Germany). All participants, who were instructed to close their eyes and relax but not fall asleep, were scanned. Resting-state fMRI data were obtained using a Gradient Echo type Echo Planar Imaging (GRE-EPI) sequence: repetition time (TR) = 2000 ms, echo time (TE) = 30ms, flip angle (FA) = 90°, field of view (FOV) = 220 × 220mm², slices = 32, thickness = 3 mm, interslice gap = 1 mm, and voxel size = 3.4 × 3.4 × 4mm³. High-resolution, three dimensional T1-weighted structural images were obtained using a magnetization prepared rapid acquisition gradient-echo (MPRAGE) sequence: TR = 1900 ms, TE = 2.52 ms, FA = 9°, slices = 176, FOV = 256 × 256mm², thickness = 1 mm, and voxel size = 1 × 1 × 1mm³.

All resting-state fMRI data were preprocessed using the Data Processing & Analysis of Brain Imaging toolbox (DPABI, Version 3.1) [50], a powerful tool built on the robust framework of Statistical Parametric Mapping (SPM; Version 8.0) and executed within the versatile MATLAB platform (Version 18a). This comprehensive workflow encompassed several critical steps designed to optimize data integrity and facilitate subsequent functional connectivity analyses: remove first 10 images, slice-timing, realignment, spatial normalization, nuisance signal regression, data scrubbing, spatial smoothing, and band-pass filtering.

Functional connectivity estimation

Following the process of data preprocessing, the comprehensive brain FC was established utilizing the Graph Theoretical Network Analysis (GRETNA) toolbox [51]. Power and colleagues defined a template of 264 potential functional areas, subdivided into 14 networks [52]. Upon the exclusion of five networks - auditory, sensory hand, sensory mouth, visual, and uncertain network - nine networks remained. These included the frontoparietal, cingulo-opercular, salience, dorsal attention, ventral attention, default mode, subcortical network, memory retrieval, and cerebellar network, comprising a total of 157 nodes. These nodes and networks were subsequently incorporated into the study. The time courses for each region of interest (ROI) were extracted, and the Pearson correlation coefficients between each pair of ROIs were calculated to represent the edge. Correlation coefficients were transformed into z-values by Fisher’s equation [53]. Although the potential inclusion of negative correlation in network analysis, the negative z-value was omitted from the data matrix due to the inherent ambiguity surrounding the interpretation of negative correlation [54, 55]. The final data matrix for each participant was a 157 × 157 z-matrix, with diagonal and negative values standardized to zero.

Network-based-statistics related to aggression and depression

NBS toolbox (version 1.2)(https://www.nitrc.org/projects/nbs/) was utilized to derive different brain FC related to aggression and depression. These patterns significantly predicted the variance in aggression and depression among the participants [56]. This procedure provides an enhanced ability to determine the connectivity of brain patterns formed by suprathreshold edge links associated with covariates of interest. In this analytical approach, aggression and depression were designated as the variables of interest, while age, gender, and head movement values (mean FD power) were incorporated as covariates. First, we performed t-tests at the edges of the fully-connected whole-brain network to assess the relationship between FC strength and aggression/depression score, and stored the t-values of the surviving edges under the chosen t-threshold to construct a t-matrix. Next, a random null distribution of a maximum pattern size exceeding the selected threshold was generated with more than 5000 permutations. The number of permutations with maximum pattern size greater than the empirical pattern size was standardized by the total number of permutations to estimate the p-value. The level of significance was set at 0.05 for this study. An initial t-threshold of 3.1 was employed for the analyses. The resulting connected brain patterns, these links showed significant correlation with individual variability in aggression and depression, were defined as a mask of threshold individual FC for subsequent analyses.

Statistical analysis

Demographic and correlation analysis

Prior to the execution of correlation analyses, a normality assessment was performed on all behavioral variables to ensure the validity of the statistical procedures. All analyses of correlation were performed by SPSS 26.0 software (IBM Corp, 2019). First, this study determined the relationship of the brain FC strength related to aggression with the CTQ scores and the aggression scores. The brain mask (t-matrix) associated with aggression was procured via the NBS toolbox. Thus, each participant’s matrix exhibited an identical pattern by the same edges with FC values and zero edges. Then, the Pearson correlation was used to calculate the relationship of each participant’s CTQ scores with aggression scores and brain FC strength.

Moreover, this study also ascertained the relationship of brain FC strength related to depression with the CTQ scores and the depression scores. The brain FC strength values associated with the BDI-II scores were also calculated for each participant by using the brain mask (t-matrix) obtained from the NBS toolbox. Then, the correlation of the CTQ scores with depression scores and brain FC strength were calculated.

Mediation analysis

In order to more thoroughly investigate the potential mediating role of resting-state FC in the relationship between childhood maltreatment and aggression, a mediation analysis was performed utilizing the PROCESS macro specifically designed for use with SPSS [57]. Specifically Model-4 in PROCESS macro was used for the analysis. In the model, we used childhood maltreatment as the independent variable, FC as the mediating variable, and aggression as the dependent variable. The bootstraps method with 5000 iterations was used to evaluate the significance of the mediating effect. The mediating effect is significant if the 95% confidence interval (CI) does not contain zero. In addition, to test another hypothesis that the relationship between childhood maltreatment and depression can be mediated by resting-state FC, another mediation analysis was conducted with childhood maltreatment as the independent variable, FC as the mediating variable, and depression as the dependent variable.

Results

The relationship of childhood maltreatment with aggression and depression

According to the claims made by Hair and Byrne, the data is considered to be normal if skewness is between -2 to + 2 and kurtosis is between ‐7 to + 7 [58, 59],

we ascertain that the three variables of childhood maltreatment, depression, and aggression conform to a normal distribution by integrating the theory with the corresponding graph techniques, including histograms, P-P plots, and Q-Q plots. We calculated the relationships between the childhood maltreatment scores and aggression/depression scores. The results of these correlations are displayed in Table 1. Statistical significance was established at a P value of < 0.05 (two-tailed). At this stage, no participants were excluded. The above calculations were performed by SPSS 26.0. Our findings revealed a significant correlation between childhood maltreatment scores and both aggression and depression scores.

Table 1.

Means and standard deviations of variables

	N	Mean	SD	Max	Mini	CTQ	BPAQ
CTQ	518	38.91	8.82	85	25	-	-
BPAQ	342	55.4	17.06	105	0	0.144*	-
BDI	518	7.41	6.64	40	0	0.194**	0.241**

* p < 0.05; ** p < 0.01

CTQ, Child Trauma Questionnaire; BPAQ, Buss Perry Aggression Questionnaire; BDI, Beck Depression Inventory

NBS results

To discern the components of brain FC patterns of aggression and depression related to childhood maltreatment, we initially employed the NBS to identify the brain FC linked to aggression and depression, respectively, significantly predicted the variation in aggression and depression. Next, we calculated the FC strength for each participant and Pearson correlations used to assess the relationship between brain FC strength and CTQ scores.

The results showed that there were 57 nodes and 56 edges associated with aggression. We identified the neuroanatomy of aggression. Figure 1(a) shows the visualization of all FCs related to aggression. In particular, we found that 12 of these increased FCs were positively associated with the CTQ scores. The enhanced FCs between right precuneus and right supramarginal gyrus (r = 0.125, p < 0.05), left dorsal anterior cingulate cortex (dACC, r = 0.168, p < 0.01) were positively correlated with the CTQ scores; the heightened FCs between right superior temporal gyrus (STG) and right/left ventral posterior cingulate cortex (vPCC)(r = 0.197, 0.125, p < 0.01), dorsal posterior cingulate cortex (dPCC, r = 0.171, p < 0.01), right dorsolateral prefrontal cortex (dlPFC, r = 0.129, p < 0.05) and left dACC (r = 0.127, p < 0.05) were positively correlated the CTQ scores; the heightened FCs between left dlPFC and left superior frontal gyrus (SFG)(r = 0.128, p < 0.05), right middle temporal gyrus (MTG)(r = 0.163, p < 0.01) were positively correlated with the CTQ scores; and the CTQ scores also exhibited significant positive relationship with increased FCs of right dlPFC - left inferior frontal gyrus (IFG)(r = 0.128, p < 0.05), left SFG - left inferior parietal lobule (IPL)(r = 0.126, p < 0.05), and left superior parietal lobule (SPL) - left thalamus (r = 0.168, p < 0.01). Additionally, we input the mask of aggression obtained by the NBS into the GRETNA platform to calculate the degree centre number of all nodes in the mask. Figure 1(b) shows the perspectives of the top ten nodes in degree centrality related to aggression, and Table 2 shows the top ten nodes with the highest contribution values in aggression nodes. The regions with the largest number of these connections were OFC, anterior prefrontal cortex (antPFC), STG, PCC, dACC, fusiform gyrus (FFG), angular gyrus, SFG, middle frontal gyrus(MFG).

Fig. 1.

(a) Visualization of all edges related to aggression (b) The perspectives of the top ten nodes related to aggression. (c) Mediating effects of the brain FC related to aggression on the relationship between CM and aggression. PCC = posterior cingulate cortex; INS = insula; IFG = inferior frontal cortex; ACC = anterior cingulate cortex; PFC = prefrontal cortex; PCUN = precuneus; IPL = inferior parietal lobe; FFG = fusiform gyrus; SPL = superior parietal lobe; MFG = middle frontal gyrus; OFC = orbitofrontal cortex; MTG = middle temporal gyrus; ANG = angular gyrus; SFG = superior frontal gyrus; STG = superior temporal gyrus; SMG = supramarginal gyrus; mFG = medial frontal gyrus; CLA = claustrum; THA = thalamus; PUT = putamen; SAN = salience network; DAN = dorsal attention network; DMN = default mode network; CON = cingulo-opercular network; FPN = fronto-parietal network; SBN = subcortical network; VAN = ventral attention network

Table 2.

The top ten nodes related to aggression

Node	Node_Name	Network	X	Y	Z
76	OFC_R	DMN	8.36	47.59	-15.18
102	antPFC	DMN	12.73	54.87	38.19
123	STG_R	DMN	52.16	-2.43	-16.4
91	PCC_L	DMN	-2.94	-48.79	12.87
111	dACC_L	DMN	-11.06	44.62	7.61
262	FFG_L	DAN	-42.26	-60.12	-8.85
96	ANG_R	DMN	52.04	-59.37	35.52
101	SFG_R	DMN	22.11	39.21	38.9
174	MFG_L	FPN	-43.93	1.8	45.7
92	vPCC_R	DMN	7.94	-48.37	30.57

OFC = orbitofrontal cortex; antPFC = anterior prefrontal cortex; STG = superior temporal gyrus; PCC = posterior cingulate cortex; dACC = dorsal anterior cingulate cortex; FFG = fusiform gyrus; ANG = angular gyrus; SFG = superior frontal gyrus; MFG = middle frontal gyrus; R, right; L, left; DMN = default mode network; DAN = dorsal attention network; FPN = fronto-parietal network

Next, the results indicated 37 nodes and 35 edges associated with depression. All edge sets associated with depression were significantly correlated with CTQ scores. We determined the neuroanatomy of depression in the same way. Figure 2(a) shows a visualization of all edges associated with depression. Of these, we found the CTQ scores exhibited significant positive relationship with strengthened FC between left ACC and left angular gyrus (r = 0.087, p < 0.05). Then, we used the same method to calculate the degree centrality number of all nodes in the mask of depression. Figure 2(b) shows a perspective view of the top ten nodes in terms of degree centrality associated with depression, and Table 3 shows the top ten nodes with the highest contribution value of aggressor nodes. These ten nodes are clustered in four regions with the highest number of connections, that is the putamen, caudate, insula, claustrum, MFG, thalamus, OFC and dACC.

Fig. 2.

(a) Visualization of all edges related to depression. (b) The perspectives of the top ten nodes related to depression. (c) Mediating effects of the brain FC related to depression on the relationship between childhood maltreatment and depression. PFC = prefrontal cortex; INS = insula; MFG = middle frontal gyrus; ACC = anterior cingulate cortex; IPL = inferior parietal lobe; ANG = angular gyrus; OFC = orbitofrontal cortex; PCC = posterior cingulate cortex; SFG = superior frontal gyrus; SMG = supramarginal gyrus; CLAU = claustrum; PUT = putamen; CAU = caudate; THA = thalamus; STG = superior temporal gyrus; CG = cingulate gyrus; SAN = salience network; DAN = dorsal attention network; DMN = default mode network; CON = cingulo-opercular network; FPN = fronto-parietal network; SBN = subcortical network; VAN = ventral attention network; MRN = memory retrieval network

Table 3.

The top ten nodes related to depression

Node	Node_Name	Network	X	Y	Z
231	PUT_R	SBN	28.52	0.82	4.01
52	INS_R	CON	36.73	0.78	-3.57
227	PUT_L	SBN	-21.97	7.48	-4.78
57	CLAU_L	CON	-34.37	3.29	4.19
210	MFG_R	SAN	36.89	32.35	-2.24
232	PUT_L	SBN	-31.38	-11.48	-0.3
233	CAU_R	SBN	14.98	4.94	7.24
234	THA_R	SBN	8.62	-3.57	5.76
76	OFC_R	DMN	8.36	47.59	-15.18
111	dACC_L	DMN	-11.06	44.62	7.61

PUT = putamen; INS = insula; CLAU = claustrum; MFG = middle frontal gyrus; CAU = caudate; THA = thalamus; OFC = orbitofrontal cortex; dACC = dorsal anterior cingulate cortex; R, right; L, left; SBN = subcortical network; CON = cingulo-opercular network; SAN = salience network; DMN = default mode network

In addition, Table 4 shows the number of edges within each network as well as between networks in aggression/depression.

Table 4.

The edges number of aggression and depression within and between networks

Aggression FC_networks	edges	Depression FC_networks	edges
DMN-DMN	31	CON-SBN	9
CON-DAN	5	SBN-SBN	7
FPN-SAN	3	CON-CON	4
DAN-SAN	3	DMN-DMN	3
FPN-FPN	2	SAN-SBN	3
FPN-SBN	2	DMN-MRN	2
DAN-DAN	2	FPN-SAN	2
DAN-SBN	2	SAN-DAN	2
DMN-SBN	1	DMN-SAN	1
DMN-VAN	1	FPN-FPN	1
FPN-DAN	1	VAN-VAN	1
VAN-FPN	1
VAN-SAN	1
SBN-SBN	1

SAN = salience network; DAN = dorsal attention network; DMN = default mode network; CON = cingulo-opercular network; FPN = fronto-parietal network; SBN = subcortical network; VAN = ventral attention network; MRN = memory retrieval network

Mediation between childhood maltreatment and aggression by functional connectivity

To investigate whether childhood maltreatment could be related to aggression and depression based on different brain FC. Using mediation analysis, we used childhood maltreatment as the independent variables, the brain FC related to aggression as the mediator variables, and aggression as the dependent variable to establish mediation model. As shown in Fig. 1(c), mediation analyses indicated that brain FC related to aggression mediated the relationship between childhood maltreatment and aggression [β = 0.096, 95% confidence interval (CI) = 0.0314 to 0.0392, p < 0.05]. The presence of standardized coefficients in the path diagram indicates the covariance between childhood maltreatment and aggression.

Mediation between childhood maltreatment and depression by functional connectivity

We also examined the mediating effect between childhood maltreatment and depression using mediation model with to childhood maltreatment as the independent variable, depression as the dependent variable, and brain FC related to depression as the mediating variable. As shown in Fig. 2(c), mediation analyses indicated that brain FC related to depression mediated the relationship between childhood maltreatment and depression [β = 0.0316, 95% confidence interval (CI) = 0.0039 to 0.0606, p < 0.05].

Discussion

This is the first study attempt to elucidate the common and distinct neural basis underlying the relationship between childhood maltreatment and aggression/depression using a large sample of resting-state fMRI data. First, NBS was applied to obtain FC related to aggression and depression, respectively. Then, mediation analyses were conducted to investigate the role of these FC patterns in mediating the relationship between childhood maltreatment and aggression/depression. The results demonstrated that increased FC patterns associated with aggression primarily contained connections within DMN, between CON-DAN, and increased FC patterns associated with depression mainly involved in connections within the reward system, CON, DMN, and between CON-reward system. Further comparison revealed that the CON and DMN play equally important roles in the relationship between childhood maltreatment and depression/aggression. Additionally, the attention network is a unique mechanism linking childhood maltreatment and aggression, whereas the reward system only worked in the association between childhood maltreatment and depression. The results of this study are consistent with our hypothesis.

Common mechanisms liking childhood maltreatment and aggression/depression

Among individuals who suffered childhood maltreatment, both those exhibiting aggressive behavior and those with depression symptom demonstrated significantly consistent abnormal changes in FC within DMN and CON-other regions. CON constitutes a key component of the executive control network, facilitating higher-order cognitive processes such as the self-regulation of thoughts, actions, and emotions [60, 61]. Dysfunction within the CON is associated with performance deficits on response inhibition, self-control and emotion regulation among individuals with aggression and depression [62–64]. A study provided a preliminary evidence of the impact of early childhood maltreatment on neural patterns associated with response inhibition in early adolescence [65]. The inability to inhibit habitually dominant responses in selecting goal-appropriate behaviors likely leads to aggression via frustration [66]. In addition, a few studies indicated that functional disconnection within the CON in depressed patients, compared to controls, led to the dysfunctions in cognitive control and emotional regulation that were characteristic of depression [67–70].

Additionally, DMN is the most robustly identifiable networks, supporting internal attention and self-referential thinking when external demands for attention are minimal [71]. Some evidence demonstrated that early life stress is associated with abnormal in both the structure and function of the DMN [72, 73]. One possible explanation is that childhood maltreatment exacerbates unpleasant mood through increased rumination [74]. The heightened connectivity of the DMN, which is oriented towards internal states, may intensify the propensity for individuals to dwell on negative emotions and experiences, thereby contributing to depressive symptoms [75, 76]. Meanwhile, alterations in DMN connectivity may also reflect dimension-specific changes in (affective) self-referential processes, indicative of uncaring and callousness traits [77]. Notably, a research review suggested that the presence of a callous and unemotional interpersonal style designates an important subgroup of antisocial and aggressive youth [78]. Consistent with prior studies, our findings also revealed that deficits in the DMN can elucidate the relationships between childhood maltreatment and aggression/depression [79–82]. To sum up, our results revealed that increased FC patterns involving the CON and the DMN serve as a common neural basis underlying the relationship between childhood maltreatment and depression/aggressive behavior.

Distinct mechanisms liking childhood maltreatment and aggression

In the present study, we emphasized that aggression-related FC involving the CON and the DAN acts as a mediator in the relationship between childhood maltreatment and aggression. The CON is implicated in high-level cognitive functions such as the control of attention and working memory [61]. DAN primarily governs top-down attention, sustained attention, and working memory [83, 84]. fMRI studies have demonstrated that the CON directs the DAN to concentrate on persistence and goal-directed stimulus responses [85, 86]. In response to an early adversity environment, some researchers discovered that maltreated individuals may exhibit diminished communication between brain regions associated with sustained attention, leading to attention deficits [87]. The distractibility and poor attention modulation may foster the emotion dysregulation believed to underlie behavioral dysregulation and aggression in maltreated children [88]. Another task-state fMRI study revealed that FC between the DMN and attention networks was positively correlated with measures of anger and aggression, aligning with a mechanism of impaired effortful control and decreased response inhibition of impulsivity [89]. Therefore, the present study revealed similar evidence, indicating that the DAN, modulated by the CON, constitutes the specific neural basis of the relationship between childhood maltreatment and aggression.

Distinct mechanisms liking childhood maltreatment and depression

In addition, our results indicated that the enhanced FC involving the CON and the reward system mediated the relationship between childhood maltreatment and depression. According to the dual-process theory of reward, the human brain comprises two distinct neural networks: the reward network, which is responsible for processing primary rewards, and the cognitive control network, which is engaged in the processing of secondary rewards [90]. A series of studies have concurred that individuals with a history of maltreatment may exhibit abnormal reward processing [20, 91, 92]. This phenomenon can be attributed to anhedonia, a principal characteristic of dysfunction within the brain’s reward system, particularly involving the frontal-striatal circuit, which includes the striatum, OFC and ACC [93, 94]. Lumley and Harkness identified that individuals with a history of emotional maltreatment reported increased levels of anhedonia [95]. Meanwhile, some researchers also suggested that widespread aberrant FC within both the reward network and the cognitive control network may represent a fundamental mechanism underlying anhedonia in the depressed patients [96]. Another study also found that abnormalities in reward processing are associated with an increased risk of depression following early adversity [34]. Similarly, the present study revealed that the reward system regulated by the CON may be a potential specific neural basis of the relationship between childhood maltreatment and depression.

Limitations

Although we utilized a large sample to confirm the common and distinct mechanisms linking childhood maltreatment to aggression and depression, further longitudinal research is required to establish causal relationships. Meanwhile, the inclusion of a larger sample that incorporates childhood maltreatment subtypes into the analysis is necessary to gain a comprehensive understanding of the associations between childhood maltreatment subtypes and aggressive and depressive outcomes. Additionally, as this study was based on self-reported questionnaires and relied on participants’ memories, there is a possibility of recall bias. Finally, the participants in this study were the youth; future studies could benefit from including a more diverse age range of participants to enhance the external validity of the study.

Author contributions

Yuan Li, Ting Zhang and Yu Mao designed the study. Yuan Li and Ting Zhang composed the set of instruments. Xin Hou was responsible for the validation part. Yuan Li performed the data analysis. Yuan Li and Ting Zhang drafted the original manuscript, and Yu Mao and Xiaoyi Chen revised and approved the final manuscript.

Funding

This research was funded by National Natural Science Foundation of China (grants 32300848), Chongqing Natural Science Foundation General Project (grants 2022NSCQ-MSX4170), Chongqing Postdoctoral Research Special Funding Project (grants 2022CQBSHTB1013), General Project of Humanities and Social Sciences Research of Chongqing Municipal Education Commission (grants 22SKGH113) and the Ministry of Education Humanities and Social Sciences Youth Project (grants 24YJCZH092).

Data availability

The data that support the findings of this study are available from the Southwest University but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of the Southwest University.

Declarations

Ethics approval and consent to participate

All participants had provided written informed consents prior experiments, and the experimental procedures approved by the ethical committee at the Southwest University (IRB NO. spy-2012-012) in accordance with the Declaration of Helsinkiwere. Clinical trial number: not applicable.

Consent for publication

Not Applicable.

Competing interests

The authors declare no competing interests.