A Comparative Analysis of Cortical Thickness in Female Youth With Non-Suicidal Self-Injury
Article information
Abstract
Objective
Non-suicidal self-injury (NSSI), prevalent among children and youth, is a risk factor for future suicidal behavior. Nonetheless, the neurobiological mechanisms of NSSI remain under-explored. This study compares the cortical thickness in female youth with NSSI and a healthy control (HC) group using brain magnetic resonance imaging (MRI) and ascertains its associations with relevant factors.
Methods
The NSSI group comprised inpatients meeting NSSI diagnostic criteria according to Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, while the HC group included youth without psychiatric symptoms. Both groups included female adolescents and young adults aged 9–24 years. Clinical data were collected for the NSSI group, and brain MRIs were obtained for both groups. Cortical thickness differences between groups were explored, adjusting for age, and correlations between cortical thickness and clinical variables in the NSSI group were analyzed.
Results
Both groups included 32 participants, with no statistically significant difference in average age (NSSI group: mean age 15.47±3.20 years; HC group: mean age 15.56±4.83 years; p>0.05). Significant intergroup differences in mean cortical thickness were observed in both the left and right hemispheres (both p=0.001). Specifically, the NSSI group exhibited significantly thinner cortical thickness in the right cingulate cortex compared to the healthy HC group (corrected p<0.05). However, correlation analysis between the cortical thickness of the right cingulate cortex and anxiety (State Anxiety Inventory for Children: r=-0.547, p=0.053) (Trait Anxiety Inventory for Children: r=-0.257, p=0.396) and Children’s Depression Inventory (r=-0.231, p=0.447) scales in the NSSI group did not reveal statistically significant correlations.
Conclusion
The study identified a noteworthy reduction in cortical thickness in the right cingulate cortex in the NSSI group.
INTRODUCTION
Non-suicidal self-injury (NSSI) involves deliberate, self-inflicted harm to one’s body without the intent of suicide. It is classified as NSSI disorder, positioned under the “Conditions for Further Study” in both the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) and its 2022 revision, the DSM-5, Text Revision (DSM-5-TR). In the DSM- 5-TR, NSSI is defined as engaging in self-injurious behavior intentionally causing bleeding, wounding, or pain to one’s body surface for five or more days over the past year [1].
Recent meta-analyses reveal a lifetime prevalence of NSSI at 22.1% among children and adolescents, with the highest rates observed in adolescents aged 10–17 years (17.2%). This prevalence gradually decreases with age to 13.4% in young adults and 5.5% in older adults [2-4]. Longitudinal studies focusing on youth NSSI indicate a prevalence peak between 15 and 17 years, followed by a decline in young adulthood [5]. Females consistently exhibit higher prevalence rates than males, consistent with findings from a study on Korean adolescents [6,7]. Owing to its adverse effects on family and interpersonal relationships, and its association with an elevated risk of future suicidal behavior, NSSI among adolescents necessitates urgent attention and intervention [8-10].
NSSI often coexists with various mental health conditions, including borderline personality disorder, mood disorders, anxiety disorders, substance use disorders, and post-traumatic stress disorder—or, it may manifest independently [11]. Adolescents engaging in NSSI commonly employ self-injury to momentarily alleviate intense negative emotions or as a form of self-punishment [12,13]. Compared to counterparts without NSSI, these adolescents exhibit more severe psychiatric symptoms, greater functional impairment, and increased difficulties in emotion regulation [14,15]. Influences from peers engaged in NSSI and social media are also significant factors [16]. Despite various proposed treatment approaches, there remains a lack of well-defined, effective interventions specifically tailored for adolescents with NSSI, highlighting the need for a deeper understanding [17,18].
Beyond these clinical and psychosocial correlates, recent neuroimaging reviews have begun to uncover the neurobiological underpinnings of NSSI and related self-harm in youth. These studies suggest changes in fronto-limbic circuits, including the amygdala, insula, and anterior cingulate cortex, as well as large-scale networks implicated in emotion regulation, salience processing, and self-referential thinking [19]. Within this broader context, structural brain magnetic resonance imaging (MRI) provides a neurobiological approach to investigate NSSI mechanisms. Previous studies comparing female adolescents meeting NSSI diagnostic criteria with healthy controls (HCs) identified smaller volumes in the insula and anterior cingulate cortex in the NSSI group [20]. Another study reported decreased gray matter volume in specific brain regions in adolescents with a history of NSSI [21]. Our previous research on subcortical volumes in female adolescents with NSSI revealed differences in the volumes of the left amygdala and thalamus compared to controls [22]. However, research into structural brain changes in individuals meeting NSSI diagnostic criteria, particularly adolescents, remains limited.
Although a few studies have examined adolescent NSSI, they have primarily focused on comparing specific brain volumes with those of HC groups. Gray-matter volume is jointly determined by cortical surface area and cortical thickness, which represent at least partly distinct and highly heritable phenotypes; accordingly, volumetric measures may obscure alterations that are specific to cortical thickness [23]. Considering that changes in cortical thickness during adolescence are associated with brain development and maturation, analyzing cortical thickness in adolescent NSSI patients is believed to offer valuable insights into the development of specific cortical regions [24]. In this context, our study expands the conventional focus on brain volume by comparing cortical thickness between female adolescents and young adults with NSSI and HCs and exploring its association with related factors.
METHODS
Participants
The NSSI group included patients who met the full DSM-5 diagnostic criteria for NSSI at the time of their brain MRI, among the patients admitted to the psychiatric ward of the Daegu Catholic University Medical Center between July 1, 2018, and December 31, 2021. The DSM-5 criteria were evaluated and confirmed by board-certified psychiatrists based on clinical interviews and review of medical records. The HC group, recruited from the community via research advertisements, reported not having a history of psychiatric diagnosis and treatment, including NSSI, during the pre-screening. Based on the findings of previous studies that NSSI is most prevalent in adolescents and young adults, particularly females, as mentioned in the introduction, female individuals aged 9 to 24 years were recruited for both groups. In both groups, those with a history of medical and neurological conditions (such as metabolic encephalopathy, seizure disorders, and brain tumors) that could affect the structure and function of the brain and those with congenital genetic disorders, schizophrenia and other psychosis, or intellectual disabilities were excluded. The study was approved by the Institutional Review Board of the Daegu Catholic University Medical Center (IRB Approval No. CR-21-156). Additionally, written consent was obtained from the participants and their parents, as appropriate for their age.
Assessment items
Clinical data
Clinical data encompassed brain MRI imaging data, psychiatric diagnosis, age of onset, severity of NSSI, history and methods of previous suicide attempts, current medications, and past and present physical illnesses for the NSSI group.
Psychometric tools
Korean version of Wechsler Adult Intelligence Scale-IV
The Korean version of Wechsler Adult Intelligence Scale-IV (K-WAIS-IV), adapted and standardized for Korea from the original WAIS-IV developed by Wechsler, is used to clinically assess the intelligence of individuals aged between 16 years and 0 months and 69 years and 11 months [25].
Korean Wechsler Intelligence Scale for Children-Third Edition
The Korean Wechsler Intelligence Scale for Children-Third Edition (K-WISC-III), a Korean adaptation and standardization of Wechsler’s WISC-III, is used to clinically assess the intelligence of children aged 6 years 0 months to 16 years and 11 months [26]. Participants recruited before December 31, 2018, were assessed using this test.
K-WISC-Fourth Edition
The K-WISC-Fourth Edition (K-WISC-IV), an adapted and standardized Korean version of the WISC-IV, is used for clinical assessment of intelligence in children aged 6 years 0 months to 16 years and 11 months [27]. Participants recruited since December 31, 2018, were assessed using this test.
Clinician-Rated-Severity-of-Non-Suicidal-Self-Injury
The Clinician-Rated-Severity-of-Non-Suicidal-Self-Injury (CRS NSSI), a scale provided by the American Psychiatric Association, enables clinicians to assess the presence and severity of a patient’s NSSI during the past year. The assessment, which is based on the number of days of NSSI and the number of methods used, as well as the need for surgical treatment after NSSI, is rated on a five-point scale [28].
State Anxiety Inventory for Children, Trait Anxiety Inventory for Children
Anxiety was assessed using the State Anxiety Inventory for Children (SAIC), a self-report measure of temporary emotional states of tension, fear, and worry, and the Trait Anxiety Inventory for Children (TAIC), a self-report measure of the individual’s inherent predisposition to anxiety. Both were adapted from Spielberger’s State-Trait Anxiety Inventory for Children. Each test comprises 20 items rated on a three-point scale, with total scores calculated [29].
Children’s Depression Inventory
The Children’s Depression Inventory (CDI), developed by Kovacs in 1977 and adapted into Korean by Cho and Lee [30], assesses depressive symptoms and related emotions in children aged 7 to 17. It comprises 27 items rated on a three-point scale, with total scores calculated. The clinical cutoff score is 21.
Edinburgh Handedness Inventory
The Edinburgh Handedness Inventory, comprising 10 items, is a quantitative assessment tool for hand usage, determining which hand is predominantly used in everyday situations [31].
Radiological data
MRI image acquisition
All participants underwent scanning using the 3.0T Signa Architect MRI scanner (GE Healthcare) at the Daegu Catholic University Medical Center. High-resolution structural images were acquired using a 3D T1-weighted BRAVO sequence with the following parameters: repetition time=6.9 ms, echo time=2.8 ms, flip angle=10°, field of view=240×240 mm, matrix size=240×240, and slice thickness=1 mm. All images were examined by a radiologist for abnormalities; the review did not yield any significant findings.
Image processing and statistical analysis
Brain imaging data were automatically pre-processed using the FreeSurfer software (Harvard University). The inputted images were subjected to motion correction and normalization, followed by alignment in the Talairach space to remove non-brain structures such as eyes, skull, and neck. Subsequently, signal intensity standardization, triangulation, and flattening were performed to generate smooth surfaces. Through a dilation process, white-matter surfaces and pial surfaces were extracted and transformed to minimize distortion and conform to a spherical standard space. This process delineated the boundaries between gray matter and white matter, based on which cortical thickness values for various brain regions were calculated [32,33].
To evaluate potential confounding variables in the comparison of each area, the average ages of the two groups were compared. Statistical analysis of data was performed using the IBM SPSS Statistics software for Windows, version 25.0 (IBM Corp.). Independent sample t-tests were conducted to compare the mean age and intelligence quotient (IQ) between the NSSI and HC groups. Analysis of cortical thickness values was undertaken using a General Linear Model in FreeSurfer, conducted independently for the left and right hemispheres, with age recorded as a covariate. To identify regions exhibiting significant between-group differences in cortical thickness while controlling for multiple comparisons, we applied a cluster-level correction set at p<0.05 using a Monte Carlo permutation with 5000 iterations. Additionally, Pearson correlation analysis was conducted to explore correlations between clinical variables and the regions manifesting intergroup differences. Considering that age is an established factor for cortical thickness, agecontrolled correlations were also examined [34-36]. The statistical significance level was set at p<0.05.
RESULTS
In total, 110 inpatients with NSSI and 34 potential HCs were initially screened. In the NSSI group, 25 patients were excluded because of structural brain abnormalities or MRI artifacts, leaving 85 eligible patients. In the HC group, one individual was excluded owing to a self-reported history of self-injurious behavior, leaving 33 eligible HCs. From the 85 eligible NSSI patients, 33 were selected so that their age distribution would closely match that of the 33 HCs, resulting in a total of 66 participants (33 in each group) who entered the cortical thickness analysis. Owing to errors during the FreeSurfer process, one participant was excluded from the analysis in each group. Consequently, 32 participants from each group (NSSI group: mean age 15.47±3.20 years; HC group: mean age 15.56±4.83 years) were included in the final analysis, resulting in a total of 64 participants. The sociodemographic and clinical characteristics of both groups are outlined in Table 1. No statistically significant difference was observed in average age between the groups (p>0.05), and both groups exclusively comprised females, right-handed participants of Asian ethnicity. The IQ was significantly higher in the HC group compared to the NSSI group (p<0.05). However, as the IQ scores of both groups fell within the normal range and there was a possibility that the NSSI group’s scores may have been influenced by psychiatric symptoms at the time of testing, IQ was not considered a relevant variable in this study.
Sociodemographic and clinical characteristics of the participants
Clinical information related to the NSSI group revealed the following: the mean age of onset for self-injurious behavior was 14.25±3.26 years, and the average duration of illness at the time of assessment was 16.55±15.30 months. The mean score on CRS NSSI was 2.56±0.66, with most patients categorized in Stages 2 (n=14) and 3 (n=15). Out of the 32 NSSI patients, 10 had a history of previous suicide attempts. The most prevalent DSM-5 diagnosis in the NSSI group was depressive disorders (n=23), followed by adjustment disorders (n=4); bipolar and related disorders (n=2); disruptive, impulse-control, and conduct disorders (n=2); and anorexia nervosa (n=1).
A statistically significant intergroup difference in average cortical thickness was observed in both hemispheres, with p-values of 0.001 for the left as well as right hemisphere (Table 2). In cortical thickness analyses performed after controlling for age as a covariate, the NSSI group exhibited significantly thinner cortical thickness in the right cingulate cortex compared to the HC group (corrected p-values <0.05) (Table 3, Figures 1 and 2). No significant intergroup difference in cortical thickness was observed in the left hemisphere.
Comparisons of average cortical thickness between NSSI group and HC group
Comparisons of cortical thickness between NSSI group and HC group
Comparisons of right cingulate cortical thickness between NSSI group and HC group. The NSSI group exhibited significantly thinner cortex in the right cingulate region compared to the HC group (corrected p<0.05). NSSI, non-suicidal self-injury; HC, healthy control.
Areas with right hemisphere cortical thickness differences between NSSI group and HC group. Red regions indicate areas where the HC group had thicker cortex than the NSSI group. No blue regions were observed, indicating that the NSSI group did not show thicker cortex than the HC group in any region. NSSI, non-suicidal self-injury; HC, healthy control.
Additionally, correlations between cortical thickness in the right cingulate cortex and clinical variables were investigated. A total of 13 patients in the NSSI group underwent SAIC, TAIC, and CDI assessments. Pearson correlation analyses were performed to examine the correlations between the right cingulate cortex thickness and related variables. A weak negative correlation was observed between the thickness of the right cingulate cortex and the SAIC scores (r=-0.547, p=0.053), albeit without reaching statistical significance. No statistically significant correlation was identified between the thickness of the right cingulate cortex and the TAIC scores (r=-0.257, p=0.396) or CDI scores (r=-0.231, p=0.447) (Table 4 and Figure 3). Additionally, the age-controlled partial correlation analyses did not reveal any statistically significant correlations between the right cingulate cortex thickness and the SAIC (r=-0.550, p=0.064), TAIC (r=-0.258, p=0.419), or CDI (r=-0.190, p=0.555) scores.
Correlation between cortical thickness of right cingulate and clinical self report scales in non-suicidal self-injury group
Scatterplot between cortical thickness of the right cingulate and clinical self-report scales in the non-suicidal self-injury group. A weak negative correlation was observed between cortical thickness and State Anxiety Inventory for Children (SAIC) scores (r=-0.547, p=0.053), although this was not statistically significant. No significant correlations were found with Trait Anxiety Inventory for Children (TAIC) or Children’s Depression Inventory (CDI) scores.
DISCUSSION
This study was conducted to investigate the structural differences in the cortical thickness of the brain between female youth diagnosed with NSSI as per DSM-5 and a HC group, and to determine the association between these structural differences and the clinical characteristics of the NSSI group. The analysis revealed thinner right cingulate cortex areas in the NSSI group compared to the HCs. To the best of our knowledge, this study has the largest sample size among structural brain imaging studies examining groups meeting the DSM-5 criteria for NSSI and therefore may provide important information on the neurobiological characteristics of youth with NSSI, although these alterations may also relate to broader vulnerability to psychopathology.
Both the NSSI and HC groups comprised female righthanded participants with no significant difference in average age. While the HC group exhibited higher overall IQ scores, this difference was not considered a confounding variable because of the potential impact of the NSSI group’s psychiatric symptoms on their test performance. The mean age of selfharm onset for the NSSI group (14.25±3.26 years) aligned with previous research findings [3]. Additionally, 31.25% of the NSSI group reported a history of suicide attempts, consistent with the 14%–70% lifetime prevalence observed in adolescent NSSI patients [11]. The majority of DSM-5 diagnoses in the NSSI group were mood-related, indicating a homogenous profile among the female youth participants.
This study compared cortical thickness differences between the NSSI and HC groups. Previous studies on patients with various psychiatric disorders, such as depression, schizophrenia, and bipolar disorder, have reported abnormalities in cortical thickness [37-39]. While the brain cortical structure is typically quantified by surface area, cortical thickness, and gray-matter volume, previous studies on NSSI patients predominantly focused on gray-matter volume [19]. However, despite high heritability in cortical thickness and surface area, there is no known correlation between these two measures, suggesting that cortical volume measurements might be influenced by unrelated genetic factors [23]. Hence, this study specifically compared the cortical thickness in NSSI patients—a novel approach in this research area. Our findings align with those of previous brain imaging studies in NSSI patients, which have shown involvement of specific brain areas. Structural brain imaging studies of NSSI adolescents have revealed reduced volumes of the insula and anterior cingulate cortex compared to HCs [20]. Functional MRI studies have shown increased neural activation in the amygdala, hippocampus, and anterior cingulate cortex in NSSI adolescents compared to HCs, indicating heightened activity in these regions [40,41]. Additionally, increased neural activity, specifically in the cingulate cortex, has been observed in NSSI patients. Our study builds upon these findings by employing a novel approach to analyze cortical thickness, providing further neurobiological evidence that cingulate cortex alterations may be involved in NSSI. The cingulate cortex, situated within the brain’s limbic lobe, encompasses various regions, including the anterior, retro-splenial, mid, and posterior cingulate cortex [42]. Of these, the anterior cingulate cortex plays a central role in emotional and behavioral regulation, serving a crucial function in conflict resolution and contributing to the processing and response to emotional stimuli as part of the salience network with the insula [43-47]. The observed reduced thickness in the cingulate cortex of NSSI patients thus suggests a potential link to difficulties in regulating negative emotions and impulsivity, and may be one neural correlate of the pathophysiological processes associated with NSSI [48].
Additional correlation analyses between SAIC, TAIC, and CDI scores and the cortical thickness of the right cingulate cortex in the NSSI group did not yield statistically significant results. However, it is well-documented that adolescents with NSSI experience greater anxiety and depression compared to their non-NSSI counterparts [49]. Additionally, they use self-injury as a tool to cope with these negative emotions [11]. Given the cingulate cortex’s involvement in processing conflict situations and emotional regulation, investigating its correlation with depression and anxiety in adolescents with NSSI is considered significant. Future studies with larger sample sizes may provide more substantial findings [43,45,46].
This study has a few limitations. First, the majority of the participants were diagnosed with mood disorders, particularly depression (22 out of 32, 68.8%). While the study population was relatively homogenous, it is noteworthy that previous research has shown that adults with depression exhibit reduced cingulate cortex thickness compared to HCs. This raises the possibility that the observed thinning of the right cingulate cortex may be influenced by the high prevalence of depressive disorders in our sample, rather than being specific to NSSI [37]. Future research should aim to recruit a more diverse sample encompassing a broader range of conditions. Additionally, it should compare NSSI patients with non-NSSI patients in addition to HCs to better isolate the specific neuroanatomical characteristics associated with NSSI. Second, although this study had a larger sample size compared to previous NSSI research, it focused exclusively on female youth, who represent the group with the highest prevalence of NSSI. This narrow focus limits the generalizability of our findings to male and adult NSSI patients. This will have to be addressed in future research by expanding the study population to include both sexes and various age groups. Third, a few of the clinical variables, namely SAIC, TAIC, and CDI, were only assessed for a subset of the NSSI group. Although partial correlation analysis was performed in this study, adjusted for age, which is known to correlate with cortical thickness, the partial administration of these tests raises concerns about potential bias and confounding factors. Given that the correlation between SAIC and target cortex thickness was statistically significant (p=0.053) despite a small sample size, further studies with larger sample sizes and comprehensive clinical assessments could provide more significant results. Finally, while participants were recruited based on the NSSI diagnostic criteria, clinical characteristics such as the onset of self-injurious behavior and duration of illness at the time of admission were diverse, resulting in high clinical heterogeneity. If future studies consider these clinical characteristics in the analysis process, they will be able to control these characteristics or find correlations according to clinical characteristics.
Despite these limitations, this study is significant because it contributes to the limited body of brain imaging research specifically focused on NSSI. Notably, it is the first of its kind to compare cortical thickness in a large sample size. Furthermore, in the Korean research context, there is a paucity of studies investigating the neurobiological mechanisms of NSSI in adolescents through brain imaging. Therefore, this study’s findings are expected to provide valuable insights and augment the existing evidence on the neurobiological underpinnings of NSSI.
This study analyzed the cortical thickness differences between NSSI patients and HCs, revealing a significant reduction in cortical thickness in the right cingulate cortex of the NSSI group. These findings suggest that alterations in the right cingulate cortex may be a potential neurobiological correlate of NSSI, possibly reflecting difficulties in emotion regulation in female youth. Given the considerable prevalence and severity of NSSI in adolescents, it is necessary to pursue further research with more extensive samples and rigorously controlled clinical factors to ascertain the structural differences in the brains of patients with NSSI.
Notes
Availability of Data and Material
The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.
Conflicts of Interest
The authors have no potential conflicts of interest to disclose.
Author Contributions
Conceptualization: all authors. Data curation: Hyanghee Jeong, Jung Yeon Moon. Formal analysis: Hyanghee Jeong, Sujin Bae. Funding acquisition: Tae Young Choi. Investigation: Hyanghee Jeong, Jung Yeon Moon. Methodology: Tae Young Choi, Sujin Bae. Project administration: Tae Young Choi. Supervision: Tae Young Choi. Visualization: Hyanghee Jeong, Sujin Bae. Writing—original draft: Hyanghee Jeong, Sujin Bae. Writing—review & editing: all authors.
Funding Statement
None
Acknowledgments
This work was supported under the framework of international cooperation program managed by the National Research Foundation of Korea (NRF-2021R1F1A1059258).