Characteristics of white matter structural connectivity in healthy adults with childhood maltreatment

ABSTRACT

Background: Childhood maltreatment (CM) is a common psychological stressor associated with multiple mental disorders. While CM is associated with vulnerability to depression and anxiety, little is known about the specific mechanism underlying this relationship.

Objective: This study aimed to investigate the white matter (WM) of healthy adults with CM and their relationships with depression and anxiety to provide biological evidence for the development of mental disorders in subjects with childhood trauma.

Methods: The CM group included 40 healthy adults with CM. The non-CM group included 40 healthy adults without CM. Diffusion tensor imaging (DTI) data were collected, and tract-based spatial statistics (TBSS) were applied to the whole brain to assess WM differences between the two groups; post-hoc fibre tractography was used to characterise the developmental differences; and mediation analysis was used to assess the relationships among the Child Trauma Questionnaire (CTQ) results, DTI indices, and depression and anxiety scores.

Results: Relative to the non-CM group, the CM group revealed significantly lower fractional anisotropy (FA) in the right posterior corona radiata (PCR-R), right anterior corona radiata (ACR-R), left super corona radiata (SCR-L), anterior thalamic radiation (ATR), and right posterior limb of the internal capsule (PLIC-R). Additionally, shorter fibre bundles passed through the PCR-R, ACR-R, and ATR in the CM group compared with the non-CM group. Besides, the length of the ACR-R mediated the relationship between CM and trait anxiety.

Conclusions: The alteration of white matter microstructure associated with childhood trauma in healthy adults may reflect biomarkers of childhood trauma. Besides, an alteration of WM microstructure in healthy adults with CM mediates the association between CM and trait anxiety, which may represent the vulnerability to developing mental disorders after childhood trauma experiences.

HIGHLIGHTS

• In this paper, we found specific alterations associated with CM in healthy adults, which may mediate the relationship between childhood trauma and trait anxiety in later life.

Antecedentes: El maltrato infantil (MI) es un estresor psicológico común asociado con múltiples trastornos mentales. Si bien el MI está asociado con la vulnerabilidad a la depresión y la ansiedad, se sabe poco sobre el mecanismo específico que subyace a esta relación.

Objetivo: Este estudio tuvo como objetivo investigar la sustancia blanca (SB) de adultos sanos con MI y sus relaciones con la depresión y la ansiedad para proporcionar evidencia biológica del desarrollo de trastornos mentales en sujetos con trauma infantil.

Métodos: El grupo MI incluyó 40 adultos sanos con MI. El grupo NO-MI incluyó a 40 adultos sanos sin MI. Se recopilaron datos de imágenes de tensor de difusión (DTI, por sus siglas en inglés) y se aplicaron estadísticas espaciales basadas en tracto (TBSS, por sus siglas en inglés) a todo el cerebro para evaluar las diferencias de MB entre los dos grupos; se utilizó tractografía de fibra post-hoc para caracterizar las diferencias de desarrollo; y se utilizó un análisis de mediación para evaluar las relaciones entre los resultados del Cuestionario de trauma infantil (CTQ, por sus siglas en inglés), los índices DTI y las puntuaciones de depresión y ansiedad.

Resultados: En relación con el grupo sin MI, el grupo MI reveló una anisotropía fraccional (AF) significativamente menor en la corona radiata posterior derecha (PCR-R, por sus siglas en inglés), la corona radiata anterior derecha (ACR-R, por sus siglas en inglés), la corona radiata izquierda superior (SCR-R-L, por sus siglas en inglés), radiación talámica anterior (ATR, por sus siglas en inglés) y extremidad posterior derecha de la cápsula interna (PLIC-R, por sus siglas en inglés). Además, los grupos de fibra más cortos pasaron por PCR-R, y ACR-R en el grupo MI en comparación con el grupo sin MI. Además, la longitud del ACR-R medió la relación entre MI y rasgos de ansiedad.

Conclusiones: La alteración de la microestructura de la sustancia blanca asociada con trauma infantil en adultos sanos puede reflejar biomarcadores de trauma infantil. Además, una alteración de la microestructura de SB en adultos sanos con MI media la asociación entre MI y ansiedad, lo que puede representar la vulnerabilidad a desarrollar trastornos mentales después de experiencias traumáticas infantiles.

背景：童年期虐待 (CM) 是与多种精神障碍相关的常见心理应激源。 虽然 CM 与抑郁和焦虑易感性有关，人们对这种关系背后的具体机制知之甚少。

目的：本研究旨在考查有 CM 的健康成人的白质 (WM) 及其与抑郁和焦虑的关系，为具有童年期创伤者精神障碍的发展提供生物学证据。

方法：CM 组包括 40 名有 CM 的健康成人。 非 CM 组包括 40 名没有 CM 的健康成年人。 收集弥散张量成像 (DTI) 数据，并在全脑使用基于束的空间统计 (TBSS) ，以评估两组之间的 WM 差异； 使用事后纤维束成像表征发育差异； 使用中介分析评估儿童创伤问卷 (CTQ) 结果、DTI 指数以及抑郁和焦虑评分之间的关系。

结果：相对于非 CM 组，CM 组在右后辐射冠 (PCR-R)、右前辐射冠 (ACR-R)、左超辐射冠 (SCR- L)、前丘脑辐射 (ATR) 和右后肢内囊 (PLIC-R)显著更低。 此外，与非 CM 组相比，CM 组中通过 PCR-R、ACR-R 和 ATR的纤维束更短。 此外，ACR-R的长度中介了CM与特质焦虑之间的关系。

结论：健康成人中童年期创伤相关的白质微观结构改变可能是反映了童年创伤的生物标志物。 此外，有 CM 的健康成人 WM 微观结构的改变中介了 CM 与特质焦虑之间的关联，这可能代表了在童年期创伤经历后发展为精神障碍的易感性。

KEYWORDS:

• Childhood maltreatment
• diffusion tensor imaging
• white matter
• stress
• depression
• anxiety

PALABRAS LLAVE:

• Maltrato infantil
• Imágenes de tensor de difusión
• Sustancia blanca
• Estrés
• depresión
• ansiedad

关键词:

• 童年期虐待
• 扩散张量成像
• 白质
• 压力
• 抑郁
• 焦虑

1. Introduction

Childhood maltreatment (CM), including physical, sexual, and emotional abuse, bullying, neglect, and social deprivation, is linked to an increased risk of physical and mental health problems in adulthood and has increasingly become a major social public health problem of widespread global concern (McCrory et al., 2017). Large-sample epidemiological surveys showed that 4–16% of children worldwide experienced physical abuse, and 10% experienced neglect or emotional abuse every year. CM is recognised as a significant risk factor for stress-related psychiatric disorders (e.g. post-traumatic stress disorder, major depressive disorder, and anxiety), even throughout their lifespan (DeRosse et al., 2020; Hart et al., 2018). While childhood trauma is associated with vulnerability to psychopathology, little is known about the specific mechanisms mediating the association between CM and mood disorders.

According to the kindling hypothesis, stress may bring lasting changes to the neurobiological system, such as neurotransmission, gene expression, and even changes in the neurological microstructure. These persistent alterations may leave long-lasting neurobiological markers that will generate increased vulnerabilities to stress in turn (Post, 1992; Zhong et al., 2020). Researchers suggested that these biological vulnerabilities were relatively stable, which is an important predictive factor of mental illness. However, due to the fact that some individuals are more likely to develop mental illness after stressful events, whereas others could recover quickly and show resilience to stress, mental illness may be the result of the interaction between vulnerabilities and low resilience (Goltermann et al., 2022; Teicher et al., 2016). Healthy individuals who have experienced stressful events in early life might form a compensatory neural mechanism associated with resilience to protect individuals from mental illness (Winter et al., 2022).

As a serious early life negative event, CM may increase vulnerability to life stress and lower the threshold required to trigger an affective episode (Carrion et al., 2007; Post et al., 2001). A growing number of reproducible findings suggested that child maltreatment was associated with brain structural and functional alterations in frontal limbic system. The most consistent finding is the alteration in the hippocampus. For instance, a meta-analysis confirmed that childhood maltreatment experiences were associated with reduced hippocampal volume (Riem et al., 2015). A longitudinal study of healthy participants also demonstrated that CM was associated with remarkable functional and structural changes, even decades later in adulthood. And importantly, limbic hyperreactivity and reduction in the hippocampal volume might be mediators between childhood experiences of adversity and the development of mood disorders (Dannlowski et al., 2012). Besides, evidence from clinical studies also suggested that childhood maltreatment was related to altered hippocampal volume. For example, depressed individuals with (vs. without) a history of abuse showed significantly reduced hippocampal volume. Meta-analysis highlighted smaller hippocampus volumes in PTSD patients with (vs. without) childhood trauma (O'Doherty et al., 2015). Another reasonably consistent finding is the alteration in the amygdala. In adults without psychopathology, enhanced amygdala activation was observed in the maltreated group than in comparison group (Jedd et al., 2015; Quide et al., 2021). The ROI-based meta-analysis found that paediatric PTSD with childhood trauma was associated with smaller amygdala volume (Kribakaran et al., 2020). Depressed individuals with a history of abuse showed increased connectivity between the amygdala and dorsolateral frontal area relative to those without abuse (Goltermann et al.). Thus, alterations in the hippocampus and amygdala might not be specific to individuals with a particular mental illness. but are neurobiological markers of childhood trauma and risk factors for the development of mental illness.

Longitudinal studies have shown that the WM size increased linearly with age, most pronounced between early childhood and adolescence (Giedd et al., 1999; Sowell et al., 2003). However, childhood trauma or stress may disrupt the process of WM development, leading to mental disorders. Using diffusion tensor imaging (DTI), an advanced noninvasive technique for studying changes in the WM microstructural organisation of major neuronal fibre pathways (Alexander & Barker, 2005; Basser, 1995; Beaulieu, 2002; Winston, 2012). Previous work found the alteration of the corpus callosum was associated with CM in clinical and non-clinical samples. A meta-analysis also reported a lower FA in the corpus callosum for children with PTSD and adult-onset PTSD with (vs. without) childhood trauma (Siehl et al., 2018). The first differences of white matter structure in the corpus callosum, left cingulate, left radiation crown, upper longitudinal tract, and left anterior thalamic radiation areas were significantly lower in schizophrenia within CM than without CM (Poletti et al., 2015). Although clinical studies have provided much evidence for CM as a vulnerability factor for psychiatric disorders, it is challenging to infer whether limbic alterations following CM are disease- or traumatic stress-related when studied among the clinical sample (Dannlowski et al., 2012). Besides, as mentioned above, healthy individuals who have experienced early trauma might not develop mental illness and form a compensatory associated with resilience, but the CM-related biomarkers may still exist. Therefore, the use of healthy human subjects is important for understanding the possible long-term effects of childhood abuse on the brain and its association with susceptibility to mental and psychological problems.

One recent study using region of interest (ROI) analysis in healthy adults revealed that CM was associated with WM alterations in the corpus callosum, corona radiata, and uncinate fasciculus (McCrory et al., 2017). Additionally, a voxel-based morphometry (VBM) study also concluded that CM was associated with reduced fractional anisotropy (FA) of the left fusiform occipital gyrus in healthy participants (Lu et al., 2013). Indeed, WM disruptions observed in healthy people exposed to CM may be associated with an increased vulnerability to depression and anxiety. For example, a longitudinal study found that WM alterations at baseline, specifically in the left superior longitudinal fasciculi, right superior longitudinal fasciculi, and right cingulum bundle projecting to the hippocampus, were associated with an increased vulnerability to unipolar depression and/or substance abuse (Huang et al., 2012). Moreover, it has been shown that a reduced FA in the posterior cingulum was related to the effects of physical neglect during childhood on trait anxiety (Tendolkar et al., 2018). In a structural MRI study, researchers also demonstrated that reduced hippocampal and medial prefrontal grey matter mediated the association between reported CM and trait anxiety in healthy adults (Gorka et al., 2014). Although there is evidence that trauma or stress in early life may disrupt the development of WM, it is unclear which fibre connections are affected, whether they are concentrated in the frontal limbic pathway, or whether WM alterations mediate the association between CM and emotional/cognitive change. In addition, the analytical methods were different in these studies. ROI-analysis is focused mainly on the selection of specific areas that may not be very representative of fibre delineations at the macrostructural level (Zhang et al., 2014). However, the main function of white matter is information transmission and communication, and it acts as a connection between the structures of cerebral grey matter in whole brain. Besides, although previous studies have found that CM was associated with fronto-limbic regions, findings were inconsistent in ROI studies. For example, ROI analysis with healthy adults revealed that CM was associated with WM alterations in the corpus callosum, corona radiata, and uncinate fasciculus (McCarthy-Jones et al., 2018). Another ROI analysis subsequently identified significant decreases in FA (p < .05) in the right parietal white matter area as well as in the right prefrontal, bilateral premotor, bilateral orbitofrontal, and temporal (Yamada et al., 2019). Thus, the ROI analysis might limit the results in the ‘interest region’, which might ignore the other significant brain regions and has spatial bias between different studies (Fox, 1991). The Whole-brain analysis could comprehensively explore all the altered brain regions associated with CM and enable a better understanding of fibre delineations at the macrostructural level. Thus, we used whole-brain TBSS analysis in our study (Zhang et al., 2014).

Tract-based spatial statistics (TBSS) is an analytical method developed specifically for DTI data (Smith et al., 2006). TBSS is carried out by estimating a ‘group average FA skeleton’ wherein the analysis is restricted to central areas of major WM-brain volumes (Zhang et al., 2014). However, previous TBSS analysis has failed to reveal a commissural connection to the lateral region of the brain, possibly because a large number of projections and long associated fibres were beyond the reconstruction (Gan et al., 2017).

Thus, in the current study, we used TBSS and post-hoc ROI tractography to examine the potentially altered WM microstructure in healthy adults with CM. We examined four diffusivity measures [FA, mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD)] within major WM pathways by TBSS to explore whether changes of large-scale brain systems are detectable in the brains of participants with CM. We used post-hoc ROI tractography to examine the anatomical connectivity via association and commissural fibres, which have not been specifically investigated previously. In particular, we tested whether CM is associated with emotional/cognitive by mediating changes in the WM. We used the Beck Depression Inventory (BDI) and the State-Trait Anxiety Inventory (STAI) to assess the depression and anxiety levels as well as the Cognitive Emotion Regulation Questionnaire (CERQ) to assess the cognitive emotion regulation strategies of the participants.

2. Methods

2.1. Participants

A total of 83 healthy young-adult participants were recruited from two universities through posters and advertisements. All participants were (1) over 18 years old; (2) right-handed; (3) able to understand the research content, (4) willing to participate in and complete the whole experiment, and (5) of Han ancestry. Two experienced psychiatrists performed structured clinical interviews based on the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) to conduct a psychiatric evaluation of all subjects. Exclusion criteria were as follows: (1) a prior DSM-IV-TR Axis I disorder; (2) a history of alcohol/substance abuse or alcohol/substance dependence; (3) a history of psychotropic medication use or psychotherapy; (4) a history of somatic disease that may potentially affect the brain, and (5) contraindication for MRI. This study was approved by the Ethics Committee of the Second Xiangya Hospital of Central South University (Code:025), and all participants provided written informed consent.

2.2. Questionnaires

2.2.1. Child Trauma Questionnaire (CTQ)

The CTQ was used to assess CM in the participants. This questionnaire is a 28-item self-report questionnaire, and the subscales included sexual, physical, and emotional abuse, and physical and emotional neglect (Bernstein et al., 1997; Bernstein et al., 2003). Each subscale contained five items, and another three items were used to validate the evaluation. The CTQ uses a 5-level rating: 1 point, never; 2 points, occasionally; 3 points, sometimes; 4 points, frequently; 5 points, always. Each subscale was scored from 5 to 25 points, and the total score ranged from 25 to 125 points. Individuals were considered to have experienced childhood trauma when they achieved cut-off scores on at least one of the subscales according to Bernstein et al. (1997) and Zhong et al. (2020) (emotional abuse ≥ 12; physical abuse ≥10; sexual abuse ≥ 8; emotional neglect ≥ 15; physical neglect ≥10). Individuals who did not meet the cut-off scores of any subscale of CTQ were divided into the non-CM group (Bernstein et al., 2003; Zhong et al., 2020). Previous studies revealed that the CTQ had good reliability and validity (Dovran et al., 2013; Gerdner & Allgulander, 2009; He et al., 2019). The Cronbach’s alpha coefficient of the total Chinese CTQ score in this study was 0.72. According to the CTQ criteria, 83 subjects were divided into the CM group (n = 42) and the non-CM group (n = 41). The demographic and clinical characteristics of the subjects are summarised in Table 1.

Table 1. Demographic and psychometric characteristics (Mean ± SD).

Characteristic	CM (n = 40)	Non-CM (n = 40)	χ^2/t	p
Sex (M/F)	20/20	17/23	χ^2 = .429	.512
Age, years	20.92 ± 2.64	21.50 ± 2.22	t = −1.04	.297
Education, years	14.74 ± 1.75	14.83 ± 1.81	t = −0.21	.828
BDI	6.83 ± 7.35	2.52 ± 3.27	t = 3.24	.002
SAI	39.59 ± 6.65	34.58 ± 4.95	3.77	.000
TAI	43.89 ± 7.53	35.95 ± 5.06	5.42	.000
CERQ	109.92 ± 12.36	107.07 ± 9.16	1.159	.250
CTQ scores
Total	45.59 ± 10.42	31.30 ± 5.15	7.53	.000
Emotional abuse	8.37 ± 4.06	6.10 ± 1.43	3.25	.002
Physical abuse	6.58 ± 2.14	5.20 ± 0.52	3.86	.000
Sexual abuse	6.37 ± 1.99	5.20 ± 0.51	3.50	.001
Emotional neglect	13.38 ± 5.07	8.35 ± 3.06	5.21	.000
Physical neglect	11.05 ± 3.13	6.45 ± 1.82	7.88	.000

M/F, number of males/number of females; BDI, Beck Depression Inventory; CTQ, Childhood Trauma Questionnaire. Means (all variables except sex) are reported with standard deviations.

2.2.2. Beck Depression Inventory (BDI)

The Chinese version of the BDI was used to assess depression severity in the participants. The BDI is a 21-item self-report scale, which is widely used worldwide. The subjects were assessed for the severity of their depressive symptoms over the past two weeks. Each item was rated as 0–3 on a 4-point scale. The BDI total score range is from 0 to 63 points, with higher scores indicating more severe symptoms. The Chinese version of the BDI has good reliability (Wu & Huang, 2014). The BDI also exhibited good internal consistency (Cronbach’s α = 0.83) in the present study.

2.2.3. State-Trait Anxiety Inventory (STAI)

The Chinese version of the STAI was used to assess anxiety severity in the participants. The widely used self-reported STAI (Bados et al., 2010) contains two parts: State Anxiety Inventory (SAI) and Trait Anxiety Inventory (TAI). Both the state anxiety and trait anxiety questionnaires contained 20 items, answered on a scale of 1–4, with a higher score indicating more severe anxiety symptoms. The STAI has a high internal consistency and good test-retest reliability (Fountoulakis et al., 2006; Kvaal et al., 2005; Ma et al., 2013). In the present study, the STAI had a good internal consistency (Cronbach’s α: state anxiety = 0.89; trait anxiety = 0.84).

2.2.4. Cognitive Emotion Regulation Questionnaire (CERQ)

The CERQ was used to assess individual’s cognitive emotion regulation strategies or coping styles after experiencing negative life events (Garnefski et al., 2004). The CERQ is a 36-item self-report scale, and each item was rated as 1–4 on a 4-level scale. The Chinese version of the CERQ has good reliability (Zhu et al., 2008), and Cronbach’s α in this study was 0.88.

2.3. DTI acquisition parameters

MRI was performed using a standard head coil with a 3.0T Siemens Magnetom Skyra scanner (Siemens Healthineers, Erlangen, Germany). The researcher explained the experiment to the subjects and made them familiar with the whole process of the experiment. Earplugs and foam pads were used to minimise noise and head movement. DTI data were acquired using a single-shot spin-echo-planar imaging sequence parallel to the line of the anterior-posterior commissure. The acquisition parameters were as follows: repetition time: 6400 ms; echo time: 86 ms; acquisition matrix: 128 × 128; field of view: 256 × 256 mm²; slice thickness: 2.5 mm; gap: none; and contiguous axial slices: 55. Diffusion-sensitive gradients were applied along 64 noncollinear directions (b = 1000 s/mm²), and an additional image was collected without a diffusion gradient (b0 = 0 s/mm²).

2.4. DTI preprocessing

DTI data was mainly analyzed by a brain imaging data processing and analysis software package (Functional Magnetic Resonance Imaging of the Brain Software Library, FSL 5.0) (http://www.fmrib.ox.ac.uk/fsl). Eddy current correction and head correction were performed on the DTI data using the eddy current correction function of FSL to eliminate the head movement during scanning and the deformation caused by head movement and eddy currents. Then a brain mask was obtained via the b0 image by the Brain Extraction Tool in the FSL software. Finally, the dtifit function in FSL was used to calculate the tensor and to obtain FA, MD, and other related indicators.

2.5. TBSS

TBSS is an exploratory analytical method based on image registration followed by whole-brain voxels. After aligning all individual FA images with a standard spatial template using nonlinear registration, the averaged FA images and the averaged skeleton were obtained. A minimum FA threshold of 0.20 was set to exclude peripheral tracts. Finally, the FA images of each subject after the nonlinear registration were projected onto the average FA skeleton for voxel-based multiple comparisons. Voxel-based multiple comparisons were performed using the threshold-free cluster enhancement (TFCE) method with the FSL randomisation procedure, with 5000 permutations, and taking the BDI score and the STAI score as covariates. The localisation of each significant cluster was visualised on a JHU ICBM-DTI-81 parcellation map with a JHU WM tractography atlas. The same method was applied for MD, AD, and RD.

2.6. Fibre tractography

The DTI data were reconstructed and analyzed using diffusion toolkit software (http://www.trackvis.org/), and fibre tracking was performed using TrackVIS software (http://www.trackvis.org/). The fibre bundle was reconstructed by continuous tracking using the fibre distribution algorithm (commonly known as FACT), which starts at the centre of each voxel and continues in the direction of the water diffusivity to the next appropriate voxel. The fibre propagation was terminated when the fibre direction changed by an angle ≥35°. Clusters with significant differences between the two groups in the TBSS analysis were deprojected in the native space following the back-projection procedure in TBSS (tbss_deproject script) and became seed points for tractography. Individual ROIs were loaded into TrackVIS software, and the FA, MD, AD, and RD values of the fibre bundles passing through them as well as the average fibre lengths were calculated.

2.7. Statistical analysis

All statistical analyses were carried out in SPSS 18.0. The independent-sample t-test was used to calculate the between-group difference in FA, MD, AD, RD, and fibre bundle length. The CM group data were subjected to correlational analyses to assess the relationships between the BDI, CERQ, and STAI scores and DTI indices in tractography. A single-level mediation analysis was used to investigate whether any potential regional WM alterations associated with the CTQ are related to depression or anxiety. The values showing significant associations with the CTQ score were then extracted and entered into SPSS 18.0 for mediation analyses using the PROCESS macro.

3. Results

3.1. Demographic and psychometric data

Two subjects with CM and one control (non-CM) subject were excluded from MRI analysis due to head movement, resulting in 40 subjects in the CM group and 40 in the control (non-CM) group. The demographic and clinical characteristics of the subjects are summarised in Table 1. The two groups were well-matched in age, sex, and years of education. The BDI, SAI, TAI, CERQ, and CTQ subscale scores were significantly larger in the CM group than in the control group (Table 1).

3.2. TBSS

The TBSS results showed significantly lower FA values in the CM group compared to the non-CM group among the five clusters, including the right posterior corona radiata (PCR-R), right anterior corona radiata (ACR-R), the left superior corona radiata (SCR-L), the anterior thalamic radiation (ATR), and the right posterior limb of the internal capsule (PLIC-R) (Table 2, Figure 1). No regions had a significantly greater FA value in the CM group relative to the non-CM group. There was no significant difference between the two groups in the RD, MD, or AD.

Figure 1. The white matter structures show significantly decreased fractional anisotropy (FA) in subjects with childhood maltreatment (CM) (p < .005 vs. non-CM, corrected for multiple comparisons). (A) Right posterior corona radiata (MNI x, y, z: 111, 93, 101), (B) right anterior corona radiata (MNI x, y, z: 117, 1141, 95), (C) left superior corona radiata (MNI x, y, z: 63, 129, 102), (D) anterior thalamic radiation (MNI x, y, z: 95, 122, 77), and (E) right posterior limb of internal capsule (MNI x, y, z: 105, 120, 76). Note: FA maps show sagittal, coronal, and axial views (from left to right). Green voxels represent the white matter skeleton of FA. Red–yellow voxels represent regions with significantly lower FA in subjects with CM relative to non-CM, thickened using the TBSS_fifill script implemented in FSL. FSL = Functional Magnetic Resonance Imaging of the Brain Software Library. TBSS = tract-based spatial statistics.

Table 2. MNI coordinates of regions with decreased fractional anisotropy in the CM group relative to the non-CM group

Variable	TFCE FWE-corrected p	MNI atlas coordinates			Voxel size	Tract(s) within clusters
		x	y	z
Cluster 1	.041	111	93	101	1204	Right posterior corona radiata
Cluster 2	.038	117	141	95	473	Right anterior corona radiata
Cluster 3	.038	63	129	102	344	Left superior corona radiata
Cluster 4	.044	95	122	77	300	Anterior thalamic radiation
Cluster 5	.046	105	120	76	150	Right posterior limb of internal capsule

3.3. Fibre tractography

The tractography results (Figure 2) showed that the mean FA values of fibre bundles passing through the ACR-R and the SCR-L were significantly less in the CM group (FA_ACR-R= 0.52 ± 0.02; FA_SCR-L = 0.51 ± 0.02) than in the non-CM group (FA_ACR-R = 0.53 ± 0.02, p = .012; FA_SCR-L = 0.53 ± 0.02, p = .049). The lengths of the fibre bundles passing through the PCR-R, ACR-R, and ATR were significantly shorter (p < .05) in the CM group than in the non-CM group.

Figure 2. Fibre tracking. (A) Fibre tracking passing through regions of interest (ROIs) of the right PCR, (B) Fibre tracking passing through ROIs of the right ACR, (C) Fibre tracking passing through ROIs of the left SCR, (D) Fibre tracking passing through ROIs of the ATR, and (E) Fibre tracking passing through ROIs of the right PLIC. PCR = posterior corona radiata, ACR = anterior corona radiata, SCR = superior corona radiata, ATR = anterior thalamic radiation, PLIC = posterior limb of internal capsule.

3.4. Correlation analysis

We found that the length of the fibre bundles passing through the ACR-R negatively correlated with the trait anxiety score (TAI) (r = −.419, p = .011, Figure 3). No significant correlations were observed between the BDI, SAI, or CERQ score and the DTI indices in the CM group (all p > .05). As shown in Figure 3, there are 3 participants, who seem to be outliers with short length of ACR-R in the CM group. Thus, we did additional correlation analysis and we found that no significant correlations were observed between trait anxiety score and fibre length of the ACR-R when excluding the three subjects (p = .083). Since the three subjects were within three standard deviations of the total sample, we still included the three participants in our sample.

Figure 3. Scatter plot of the right anterior corona radiata (ACR-R) fibre length and the Trait Anxiety Inventory (TAI) score.

3.5 Sensitivity analysis

Sensitivity analyses were used to test whether the primary findings (fibre length of the PCR-R, ACR-R, and ATR) would be affected by different types of CM (i.e. sexual, physical, and emotional abuse, physical and emotional neglect) (Dong et al., 2022). Besides, previous studies proposed that different dimensions of CM (i.e. threat and deprivation) might have differential effects on brain development (McLaughlin et al., 2014; McLaughlin et al., 2019). Thus, we also examined sensitivity analysis in threat and deprivation dimensions which CTQ assessed. Threats involve varying degrees of physical and sexual abuse, witnessing domestic violence, and so on, while deprivation involves various forms of neglect (McLaughlin et al., 2019). Thus threat dimension included sexual, physical, or emotional abuse, and the deprivation dimension included physical or emotional neglect (Dong et al., 2022). The sensitivity analysis results showed that the fibre length of PCR-R and ACR-R was significantly shorter (p < .05) in the CM group than in the non-CM group in most types of trauma. Our findings remained significant when applying either threat or neglect deprivation criteria in PCR-R and ACR-R fibre length (see Table S1, supplemental data).

3.5. Mediation analysis

To explore whether the ACR-R mediated the association between CM and trait anxiety, we used mediation analysis to test whether the TAI score (Y) as an outcome is mediated by the CTQ total score (X) through its effect on the ACR-R length (M). Mediation analysis revealed that the length of the ACR-R significantly mediated the relationship between the CTQ total score and the TAI score (β = 0.216, lower limit of the confidence interval = 0.024, upper limit of the confidence interval = 0.437; Figure 4).

Figure 4. Displayed is the single-level mediation analysis. The following factors were used: X = CTQ total score, M = fibre length of the right anterior corona radiata (ACR-R), Y = STAI total score. Three different paths depict the relation between the different factors. Path a: relationship between X–M, path b: relationship M–Y, path ab: mediation of relationship X–Y by M. The lines are labelled with path coefficients, and standard errors are shown in parentheses. **p < .01.

4. Discussion

In the current study, we used TBSS and post-hoc ROI tractography to explore WM alterations in healthy adults with CM. We found that self-reported CM was associated with reduced FA values in the corona radiata (PCR-R, ACR-R, and SCR-L), ATR, and PLIC regions. Our tractography data confirmed these differences, showing FA values and fibre length alterations with significant group differences. Besides, we also found that an alteration of the ACR-R length mediates the association between self-reported CM and trait anxiety in adulthood. To the best of our knowledge, this study is the first to use TBSS and post-hoc tractography to comprehensively examine WM microstructure alterations in individuals with CM. Our findings extend previous studies and may provide a basis for the formation of a mechanism from CM to affective disorder.

Consistent with previous studies, the WM change in the corona radiata and ATR of individuals with CM implicates an alteration of structure in the fronto-limbic system in mental disorders in adulthood (Hart & Rubia, 2012; Lim et al., 2020). The corona radiata is part of the limbic-thalamic circuit and includes thalamic radiation from the internal capsule to the cortex. These areas are related to top-down emotional regulation. A diffusion tensor imaging study of 147 healthy adults showed a significant lower FA values in the corona radiata regions among the CM group compared with the non-CM group (McCarthy-Jones et al., 2018). Besides, the corona radiata plays an important role in the control of attention and is part of the executive attention network (Qiu et al., 2011). Therefore, WM alteration in the corona radiata may impact attention. According to a study on adolescents, students who had experienced early neglect had significantly lower FA values in the corona radiata than those who had not experienced early neglect, and this reduction may further lead to cognitive deficits in adolescents (Hanson et al., 2013). Apart from the corona radiata, the ATR connects the thalamus with the prefrontal cortex and the occipital cortex, and it is also a part of the fronto-limbic system. This WM tract is related to emotional expression, the regulation of reward-seeking behaviours, and the regulation of emotional states (Tendolkar et al., 2018). Consistent with the results of the current study, Tendolkar et al. have reported that a reduced FA of the ATR is related to the severity of childhood physical neglect in healthy individuals (Tendolkar et al., 2018). The alteration in FA indicates a change in diffusion anisotropy, which provides exclusive information about the directionality of axons in the brain and may indicate a loss of axons and myelin. Thus, lower FA values in the corona radiata and ATR suggests the key major WM tracts connecting the fronto-limbic system are altered after CM and may further lead to cognitive deficits.

Combining TBSS and fibre tractography methods, we also found fibre length alterations in the ACR, SCR, and ATR of CM brains, relative to non-CM brains. Analysis of the length of the fibre bundle also represents a measurement of the macrostructure and provides additional information regarding the bundle geometry (Zhang et al., 2014). The measurement of the bundle length reflects the estimated range of the fibre bundle. Our findings of alteration WM and short fibre bundles in the ACR, SCR, and ATR may lead to hypo-transmission of information in the fronto-limbic system. The white matter tracts and pathways connect functional cortical, and thus it is important for cognitive processes, including the speed of information processing (Fields, 2008; Tsuda et al., 2005). Fibre length is a quantitative indicator for white matter (Beaulieu, 2002; Gao et al., 2011). Previous studies has proposed that the lengths of axonal fibres were particularly decreased in shorter, smaller-diameter myelinated regions, which might affect information processing (Behrman-Lay et al., 2015). A length-based tractography MRI study also conducted that fibre bundle lengths shortening in specific brain regions related to poorer executed function and cognitive processing speed performance among older adults (Behrman-Lay et al., 2015). Critically, the sample used in this study consisted of healthy adults who demonstrated no current or past psychiatric disorders. Thus, this structural connectivity alteration has only been found in healthy samples with vulnerability characteristics indicating that they are related to CM independent of psychiatric disease.

Moreover, we used single-level mediation analysis to investigate whether any potential regional WM alterations associated with the CTQ score are related to the impact of the CTQ score on trait anxiety. We found a significant mediation, indicating that the fibre length changes in the ACR-R may have a functional consequence of childhood maltreatment and are linked to trait anxiety, a personality dimension that is suggested to be a transdiagnostic risk factor of affective disorder (Gorka et al., 2014). Studies of the underlying neurobiology of this risk have identified structural and functional alterations in cortico-limbic brain circuits, which seem particularly sensitive to these early adverse experiences and are associated with anxiety and affective disorders (Aghamohammadi-Sereshki et al., 2021; Kaplan et al., 2016; McCrory et al., 2017). Based on previous studies, we further concluded that WM alterations within the ACR-R mediate the association between the self-reported CM and the increased expression of trait anxiety in adulthood. Although these conclusions need to be further confirmed in future studies, our findings may provide a basis for the formation of a mechanism from CM to affective disorder.

This study had several limitations that should be mentioned. First, the CM assessment in the current study relied on retrospective, self-reported measurement data that may be biased by the subject’s subjective experience and current emotional state. And the retrospective assessment tools used in this study did not take the time and duration of the trauma into account, which may be essential to clarify the impact of CM on neurodevelopment. A structured interview to assess the level of CM was required in future research. Nevertheless, the questionnaires have been proven reliable and effective tools for measuring childhood trauma, with high internal consistency and test-retest reliability (Dovran et al., 2013; Gerdner & Allgulander, 2009; He et al., 2019). Second, this study was a cross-sectional study. We assessed self-reported CM, trait anxiety, and WM microstructure simultaneously, so they are correlated in nature and cannot be shown to be causal. A combination of cross-sectional and longitudinal studies in high-risk populations can advance the underlying mechanism of the aetiology and pathophysiology of childhood trauma-related affective disorder. Third, unpredictability is an important dimension in childhood maltreatment which have lasting effects on neuro-development (McLaughlin et al., 2019). However, we didn’t add this dimension to sensitivity analysis. Fourth, we adopted a standard method to perform a voxel-based analysis of DTI data under a 3.0T field strength in 64 directions. However, it should be noted that DTI sequences generally have limitations in the interpretation of cross-fibre regions. Besides, the development of white matter is known to differ in psychiatric disorders (e.g. PTSD, MDD) and in healthy samples. However, the current study was conducted in healthy population and the results could not be extended to clinical and sub-clinical populations. Therefore, future studies should include sub-clinical or clinical samples.

5. Conclusion

The current study showed microstructural alterations in the brains for individuals with CM, predominantly in the corona radiata, ATR, and PLIC-R. In addition, we also found that the length of the ACR-R mediated the relationship between the CTQ score and individual differences in trait anxiety. Our findings together suggested that structural alterations in these regions increased affective sensitivity to stress later in life for those who experienced early adversity.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by National Science and Technology Project for Professional Basic Research [grant number 2015FY111600]; National Natural Science Foundation [grant number 81370034].