Introduction
Major depressive disorder (MDD) is a psychiatric disorder clinically characterized by changes in mood and cognition, loss of interest or pleasure, lasting at least 2 weeks, and sometimes accompanied by suicidal behavior (Dong et al., Reference Dong, Haque, Wu, Placide, Yu and Zhang2021; Zhdanava et al., Reference Zhdanava, Pilon, Ghelerter, Chow, Joshi, Lefebvre and Sheehan2021). From 2001 to 2020, the lifetime prevalence of MDD is 19% (95% CI: 0.12–0.26) globally, with a pooled 1-year prevalence of about 8% (95% CI: 0.05–0.12) (Shorey, Ng, & Wong, Reference Shorey, Ng and Wong2022). In mainland China, the lifetime prevalence and 12-month prevalence of MDD were 3.4% (95% CI: 2.9–3.9) and 2.1% (95% CI: 1.8–2.4), respectively (Huang et al., Reference Huang, Wang, Wang, Liu, Yu, Yan and Wu2019). The incidence of suicidal ideation (SI) in Chinese individuals with MDD is as high as 53.1% (95% CI: 42.4–63.4%) (Dong et al., Reference Dong, Wang, Li, Xu, Ungvari, Ng and Xiang2018). Therefore, there is a need to search for risk factors associated with SI in MDD patients, and to provide regular screening and effective intervention for suicidal behavior (Dong et al., Reference Dong, Wang, Li, Xu, Ungvari, Ng and Xiang2018; Dong et al., Reference Dong, Haque, Wu, Placide, Yu and Zhang2021). SI in MDD patients is significantly associated with many sociodemographic characteristics, such as lower educational attainment, Caucasian or African American, male sex, unemployment, and psychiatric treatment, as well as clinical characteristics, such as previous suicide attempts, younger age of onset, and glucose disturbances (Dong et al., Reference Dong, Haque, Wu, Placide, Yu and Zhang2021; Sokero et al., Reference Sokero, Melartin, Rytsala, Leskela, Lestela-Mielonen and Isometsa2003; Trivedi et al., Reference Trivedi, Morris, Wisniewski, Nierenberg, Gaynes, Kurian and Rush2013). However, the underlying neural mechanisms behind these risk factors remain unclear.
Various cortical and limbic structures, including the anterior cingulate cortex (ACC), have been implicated in MDD (Chen et al., Reference Chen, Wang, Niu, Zhong, Hu, Chen and Huang2018; Ibrahim, Kulikova, Ly, Rush, & Sherwood Brown, Reference Ibrahim, Kulikova, Ly, Rush and Sherwood Brown2022; Lemke et al., Reference Lemke, Romankiewicz, Forster, Meinert, Waltemate, Fingas and Dannlowski2022; Sindermann et al., Reference Sindermann, Redlich, Opel, Bohnlein, Dannlowski and Leehr2021). The ACC is a limbic structure that plays an important role in regulating emotion, attention, motivation, and information processing (Bliss-Moreau, Santistevan, Bennett, Moadab, & Amaral, Reference Bliss-Moreau, Santistevan, Bennett, Moadab and Amaral2021; Hu et al., Reference Hu, Zhang, Xia, Zhao, Li, Yang and Cao2021; Li et al., Reference Li, Yu, Huang, Chen, Ai, Zhou and Kuang2021; Smith, Asada, & Malenka, Reference Smith, Asada and Malenka2021). It is part of the emotional circuit known as the “anterior compartment.” This circuit also includes the amygdala and the anterior insula, which are anatomically interconnected and metabolically integrated (Ibrahim et al., Reference Ibrahim, Kulikova, Ly, Rush and Sherwood Brown2022; Mayberg et al., Reference Mayberg, Liotti, Brannan, McGinnis, Mahurin, Jerabek and Fox1999). Multiple magnetic resonance imaging (MRI) studies have reported reduced ACC volume in MDD patients (Belleau, Treadway, & Pizzagalli, Reference Belleau, Treadway and Pizzagalli2019; Mertse et al., Reference Mertse, Denier, Walther, Breit, Grosskurth, Federspiel and Bracht2022; Riva-Posse, Holtzheimer, & Mayberg, Reference Riva-Posse, Holtzheimer and Mayberg2019). This finding was confirmed by a larger study in which individuals with higher self-reported severity of depressive symptoms had reduced right-sided ACC volume, an association that was only significant in men (Ibrahim et al., Reference Ibrahim, Kulikova, Ly, Rush and Sherwood Brown2022).
In patients with MDD, ACC dysfunction is associated with SI (Lewis et al., Reference Lewis, Port, Blacker, Sonmez, Seewoo, Leffler and Croarkin2020). Reduced gray matter volume (GMV) of the ACC has been found in suicide attempters among individuals with various psychiatric disorders (Bani-Fatemi et al., Reference Bani-Fatemi, Tasmim, Graff-Guerrero, Gerretsen, Strauss, Kolla and De Luca2018). GMV of ACC may be a sensitive biomarker of SI risk in MDD patients. Abnormalities in brain structure may be associated with the development and progression of SI in MDD patients. However, the expression of GMV in the ACC of MDD patients with SI remains unclear.
A large number of previous studies have attempted to reveal the neural mechanisms behind suicidal behaviors through structural neuroimaging techniques (Chen et al., Reference Chen, Xu, Li, Shu, Zhou, Guo and Liang2024; Guo et al., Reference Guo, Jiang, Jia, Zhu, Han, Wu and Tang2023). Previous studies have shown an association between changes in GMV in the brains and SI in patients with MDD (Chen et al., Reference Chen, Xu, Li, Shu, Zhou, Guo and Liang2024; Guo et al., Reference Guo, Jiang, Jia, Zhu, Han, Wu and Tang2023). Reduced GMV in the bilateral caudate nucleus (Ho et al., Reference Ho, Teresi, Ojha, Walker, Kirshenbaum, Singh and Gotlib2021), right nucleus accumbens (Ho et al., Reference Ho, Teresi, Ojha, Walker, Kirshenbaum, Singh and Gotlib2021), dorsal striatum (Ho et al., Reference Ho, Cichocki, Gifuni, Catalina Camacho, Ordaz, Singh and Gotlib2018), lingual gyrus (Wang et al., Reference Wang, Kang, Zhang, Guo, Wang, Zong and Liu2021), right insula (Ge et al., Reference Ge, Hassel, Arnott, Davis, Harris, Zamyadi and Vila-Rodriguez2021), bilateral dorsolateral prefrontal cortex (DLPFC; R Zhang et al., Reference Zhang, Wei, Chang, Jiang, Tang and Wang2020), and right ventrolateral prefrontal cortex (R Zhang, Wei, et al., Reference Zhang, Wei, Chang, Jiang, Tang and Wang2020) was associated with SI in patients with MDD. Two comprehensive meta-analyses reported significant changes in cortical and subcortical structure in MDD patients with suicidal tendencies compared to healthy controls, including smaller total intracranial and subcortical volumes, larger ventricular volumes, smaller thalamus and globus pallidus volumes, and smaller inferior parietal lobe surface area (Campos et al., Reference Campos, Thompson, Veltman, Pozzi, van Veltzen, Jahanshad and Group2021; Renteria et al., Reference Renteria, Schmaal, Hibar, Couvy-Duchesne, Strike, Mills and Hickie2017). These findings suggest that neurobiological markers are associated with suicidal behavior in depressed patients (Chen et al., Reference Chen, Xu, Li, Shu, Zhou, Guo and Liang2024; Guo et al., Reference Guo, Jiang, Jia, Zhu, Han, Wu and Tang2023). However, no diagnostic clinical imaging markers and predictors have been identified (Yan et al., Reference Yan, Chen, Li, Castellanos, Bai, Bo and Zang2019). One of the reasons for these contradictory results is that small sample studies have low statistical power and are prone to false positive results (Button et al., Reference Button, Ioannidis, Mokrysz, Nosek, Flint, Robinson and Munafo2013; Poldrack et al., Reference Poldrack, Baker, Durnez, Gorgolewski, Matthews, Munafo and Yarkoni2017). Second, different workflows can lead to differences in the results of MRI datasets (Botvinik-Nezer et al., Reference Botvinik-Nezer, Holzmeister, Camerer, Dreber, Huber, Johannesson and Schonberg2020). Finally, physiological confounders such as head motion can also affect results (Ciric et al., Reference Ciric, Rosen, Erus, Cieslak, Adebimpe, Cook and Satterthwaite2018). The REST-meta-MDD project shared 25 research cohorts, including R-fMRI data from 1,300 MDD patients and 1,128 healthy control participants (Chen et al., Reference Chen, Lu, Li, Li, Wang, Castellanos and Yan2022). A standardized preprocessing procedure was used for each subject to minimize heterogeneity in preprocessing methods (Chen et al., Reference Chen, Lu, Li, Li, Wang, Castellanos and Yan2022; Yan, Wang, Zuo, & Zang, Reference Yan, Wang, Zuo and Zang2016). All data were corrected for head movements using the Friston-24 model as the default setting (Yan et al., Reference Yan, Cheung, Kelly, Colcombe, Craddock, Di Martino and Milham2013). The project improved statistical power by pooling data across centers while minimizing the impact of heterogeneous analysis strategies (Chen et al., Reference Chen, Lu, Li, Li, Wang, Castellanos and Yan2022).
However, there are no large, multicenter studies investigating structural brain changes in MDD patients with SI by GMV. In particular, no study has used samples from the Chinese REST-meta-MDD project (Yan et al., Reference Yan, Chen, Li, Castellanos, Bai, Bo and Zang2019), which addresses the problems of limited statistical power and analytical heterogeneity of small samples. In this study, we sought to elucidate the neurological basis of MDD patients with SI by comparing the GMV of structural MRI of ACC brain regions in MDD patients with and without SI from the REST-meta-MDD project. We further sought to determine whether these altered brain regions were risk factors for SI in MDD patients. To the best of our knowledge, similar studies have not been conducted in MDD patients in the Chinese population. We hypothesized that there is a difference in GMV of ACC between MDD patients with and without SI.
Methods
Participants
Participants in this study included 1,300 individuals diagnosed with MDD and 1,128 healthy controls. As part of the REST-meta-MDD consortium, all participants were recruited from 25 Chinese consortium members in 18 hospitals based on similar inclusion and exclusion criteria (Liu et al., Reference Liu, Li, Zhang, Sun, Li, Chen and Zhang2021; Long et al., Reference Long, Cao, Yan, Chen, Li, Castellanos and Liu2020; Yan et al., Reference Yan, Chen, Li, Castellanos, Bai, Bo and Zang2019). All individuals provided written informed consent before participating in the study. The original studies were approved by the local Institutional Review Boards, and then by the Institutional Review Board of the Institute of Psychology, Chinese Academy of Sciences for sharing deidentified and anonymized data. Consortium members provided only basic information, including diagnosis, duration of illness, sex, age, education status, and the 17-item Hamilton Depression Rating Scale (HAMD).
A total of 88 MDD patients and 38 healthy controls were excluded from the analysis. Patient inclusion criteria were as follows: (1) age over 18 and under 65 years; (2) years of education greater than 5 years; (3) meeting the criteria for MDD based on the Structured Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV) or International Classification of Diseases 10 (ICD-10); and (4) having a total score of no less than 8 on the 17-item HAMD at the time of scanning.
The exclusion criteria were as follows: (1) exclusion of patients with late-onset depression and patients in remission; (2) exclusion of subjects who lacked basic information, such as incomplete information on gender, age, and education; (3) exclusion of subjects with poor quality imaging data, including poor spatial normalization; (4) exclusion of subjects with group mask coverage of less than 90% or with a head-averaged framewise displacement of greater than 0.2 mm; and (5) exclusion of subjects from research sites with sample sizes less than 10 people (Yan et al., Reference Yan, Chen, Li, Castellanos, Bai, Bo and Zang2019).
We matched individuals with and without SI for age and sex. MDD patients were also matched with healthy controls for age and sex. Therefore, this study included 235 healthy controls and 246 MDD patients, including 123 MDD patients with and without SI each.
Clinical measures
The 17-item HAMD was used to assess the level of depression in MDD patients (Hamilton, Reference Hamilton1960). The HAMD scale consists of 17 items, 8 of which are on a 5-point level (0: absent, 4: severe) and 9 items are on a 3-point level (0: absent, 2: severe). The presence and degree of depression is determined by the total score of HAMD. The Chinese version of this scale has been shown to have good reliability and validity (Dong et al., Reference Dong, Haque, Wu, Placide, Yu and Zhang2021; Lin, Reference Lin1986; Sun, Li, Yu, & Li, Reference Sun, Li, Yu and Li2017).
SI was measured by HAMD item 3 (suicide). The item has the following alternative statements: 0 = not present, 1 = feeling that life is not worth living, 2 = wishing for or repeatedly thinking about death, 3 = having suicidal thoughts, and 4 = attempting suicide. For the purposes of this study, we defined someone had SI via a score of ≥3 on HAMD item 3. Specifically, if an individual scores 3 or 4 on HAMD item 3, he/she will be classified into the group with SI, and individuals with scores of 0, 1, and 2 will be classified into the group without SI (Ge, Jiang, Wang, Yuan, & Zhang, Reference Ge, Jiang, Wang, Yuan and Zhang2020; Li et al., Reference Li, Liu, Hou, Baldwin, Li, Cui and Sun2019; Vuorilehto et al., Reference Vuorilehto, Valtonen, Melartin, Sokero, Suominen and Isometsa2014).
Image acquisition, preprocessing, and quality control
Structural T1-weighted MRI brain scans were performed at each hospital (Yan et al., Reference Yan, Chen, Li, Castellanos, Bai, Bo and Zang2019). Images were preprocessed using DPARSF software, and image processing was performed using SPM 8 and VBM 8 toolboxes (http://dbm.neuro.unijena.de/vb) (Chao-Gan & Yu-Feng, Reference Chao-Gan and Yu-Feng2010; Yan et al., Reference Yan, Wang, Zuo and Zang2016). T1 images were normalized using template space, and then segmented into gray matter, white matter (WM), and cerebrospinal fluid. Individual native spaces were then converted to MN1 space using the Diffeomorphic Anatomical Registration Through Exponential Lie algebraic tool (Ashburner, Reference Ashburner2007). Once preprocessing was complete, quality checks were performed using the two modules called “Display one slice for all images” and “Check sample homogeneity using covariance.” The normalized images were smoothed using an 8 mm full width at half maximum Gaussian kernel.
Statistical analysis
Demographic and clinical variables were compared using the χ2 test, categorical variables using Fisher’s exact test, and continuous variables using analysis of variance. Normal distribution was assessed using the Kolmogorov–Smirnov test (p > 0.05). The Levene’s test was used to assess homogeneity of variance (p > 0.05), and the Mauchly’s test was used for the assumption of sphericity (p > 0.05).
Using the DPABI toolbox version 7.0 (Yan et al., Reference Yan, Wang, Zuo and Zang2016), GMV in the ACC was compared and analyzed using two-sample t tests with age, education, HAMD, and head movement as covariates to determine differences between MDD patients and healthy controls. In the patient group, two-sample t tests were performed with concomitant SI as the independent variable, GMV in ACC as the dependent variable, and age, education, first episode, medication status, disease duration, HAMD, and head movement as covariates. The mask was areas of interest for ACC defined by the Harvard-Oxford atlas (Carlson et al., Reference Carlson, Fang, Koster, Andrzejewski, Gilbertson, Elwell and Zuidema2022; Desikan et al., Reference Desikan, Segonne, Fischl, Quinn, Dickerson, Blacker and Killiany2006; Frazier et al., Reference Frazier, Chiu, Breeze, Makris, Lange, Kennedy and Biederman2005). In addition, we adjusted multiple testing using 1000 permutations with threshold-free cluster enhancement (TFCE) correction.
Similarly, we performed a correlation analysis between GMV in the ACC and demographic and as well as clinical variables. Brain regions were considered significant according to the Gaussian random field theory (GRF) correction (cluster-P <0.01, voxel-P <0.001). We then extracted GMV under the clusters with significant correlations and used multiple regression models to analyze the factors influencing abnormal GMV in the ACC of MDD patients with and without SI.
All statistical analyses were calculated using R version 4.3.1 (http://cran.r-project.org). Data are presented as mean ± standard deviation. The p-value was set at a two-tailed significance level ≤0.05.
Results
Demographic and clinical characteristics
Demographic information is shown in Table 1. There was a significant difference in education between MDD patients and healthy controls (p < 0.001). There were also significant differences in HAMD (p < 0.001) and head motion (p < 0.05) between MDD patients with and without SI. These factors were controlled as covariates in subsequent analyses.
Table 1. Demographics of MDD patients and healthy controls
Note: Mean ± SD; HAMD: Hamilton Depression Scale; MDD: major depressive disorder; SI: suicidal ideation.
Differences in GMV in the ACC of MDD patients with and without SI and healthy controls
As shown in Supplementary Figure S1, GMV of the right ACC was significantly smaller in MDD patients than in healthy controls (cluster 1: peak voxel X = 6, Y = 18, Z = 21; t = 4.387; cluster 2: peak voxel X = 6, Y = 15, Z = 22.5; t = 6.201). After adding age and education as covariates, one of the differences remained significant (cluster size = 104; peak voxel X = 6, Y = 15, Z = 22.5; t = 6.131). All of these results passed 1000 permutations with TFCE correction.
Furthermore, we analyzed the differences in GMV of ACC between MDD patients with and without SI. The GMV of bilateral ACC was greater in MDD patients with SI compared to MDD patients without SI (see Table 2 and Figure 1). After adding age, education, first episode, medication status, disease duration, HAMD and head motion as covariates, significant differences remained between the four clusters (see Supplementary Table S1). All of these results passed 1000 permutations with TFCE correction.
Table 2. Regions showing significant differences between GMV of MDD patients with SI and without SI
Note: BA = Brodmann area; MNI coordinates = Coordinates of primary peak locations in the Montreal Neurological Institute space; T-statistical value of peak voxel showing GMV differences between groups; 1000 permutations with TFCE corrected for multiple comparisons across space.
Figure 1. Clusters with significant differences in GMV in the ACC between MDD patients with SI and without SI based on two-sample t tests. Red and blue colors denote increased and decreased GMV. The color bars indicate the T-value (1000 permutations with TFCE correction).
Differences in the relationships between ACC GMV and demographic characteristics in MDD patients with and without SI and healthy controls
Figure 2 shows that ACC GMV was significantly correlated with age and SI in MDD patients. The specific data are presented in Supplementary Table S2. However, no correlation was found between GMV of ACC and demographic characteristics in healthy controls (all p > 0.05).
Figure 2. Correlation between GMV in the ACC and age as well as SI in MDD patients A: Significant negative correlations of a cluster (X = -6 Y = 31.5 Z = 18) in ACC and age (p < 0.05). B: Significant negative correlations of a cluster (X = 4.5 Y = 36 Z = 7.5) in ACC and age (p < 0.05). C: Significant negative correlations of a cluster (X = 4.5 Y = 24 Z = 16.5) in ACC and age (p < 0.05). D: Significant negative correlations of a cluster (X = -3 Y = 37.5 Z = -6) in ACC and age (p < 0.05). E: Clusters with significant correlation between GMV in the ACC and SI in MDD patients. The color bar represents the correlation coefficient r (adjusted by GRF, cluster-P <0.05, voxel-P <0.001).
A negative correlation between GMV in right ACC and sex was observed in MDD patients with SI (cluster size = 42, peak voxel X = 1.5, Y = −6, Z = 36; r = −0.400, adjusted by GRF, cluster-P <0.05, voxel-P <0.001).
Table 3 shows the relationship between GMV of ACC and demographic characteristics in MDD patients without SI. A negative correlation between GMV of left ACC and age (Supplementary Figure S2) and between GMV of bilateral ACC and sex was noted in MDD patients without SI (two p < 0.05).
Table 3. Associations between the GMV in the ACC and demographic characteristics in MDD patients without SI
Note: BA = Brodmann area; MNI coordinates = coordinates of primary peak locations in the Montreal Neurological Institute space; p-value for Pearson correlation; adjusted by GRF, cluster-P <0.05, voxel-P <0.001.
Further multiple regression analysis showed that in MDD patients, sex (t = −5.287, p < 0.001), age (t = −5.731, p < 0.001), and first episode (t = 2.557, p < 0.05) were factors associated with age-related GMV of the first cluster. Sex (t = −5.203, p < 0.001), age (t = −5.107, p < 0.001), first episode (t = 2.551, p < 0.05), and on medication (t = 1.978, p < 0.05) factors associated with age-related GMV of the second cluster. Sex (t = −3.248, p < 0.01), age (t = −5.284, p < 0.001), and education (t = −2.361, p < 0.05) were factors associated with age-related GMV of the third cluster. Sex (t = −5.424, p < 0.01), age (t = −5.12, p < 0.001), and first episode (t = 2.866, p < 0.05) were factors associated with age-related GMV of the fourth cluster.
Multiple regression analysis showed that among MDD patients with SI, sex (t = −5.078, p < 0.001), age (t = −2.350, p < 0.05), and education (t = −2.293, p < 0.05) were factors associated with sex-related GMV of the cluster.
In MDD patients without SI, sex (t = −5.636, p < 0.001), age (t = −5.4, p < 0.001), and first episode (t = 2.611, p < 0.05) were factors associated with age-related GMV of the cluster. Sex (t = −8.413, p < 0.001) and age (t = −3.621, p < 0.001) were factors associated with sex-related GMV of the first cluster. Sex (t = −7.622, p < 0.01), age (t = −2.355, p < 0.05), and first episode (t = 2.280, p < 0.05) were factors associated with sex-related GMV of the second cluster. Sex (t = −7.306, p < 0.001) and age (t = −2.693, p < 0.001) were factors associated with sex-related GMV of the third cluster.
In addition, in MDD patients, logistic regression was performed with the presence or absence of comorbid SI as the dependent variable. In patients with MDD, HAMD (t = 4.83, p < 0.001), medication (t = −2.593, p < 0.001), and GMV of cluster were significantly associated with SI (t = 4.418, p < 0.001; t = 2.934, p < 0.01).
Discussion
This study had three main findings as follows: (1) MDD patients with SI scored higher on HAMD compared to MDD patients without SI, (2) GMV in the ACC region was significantly smaller in MDD patients compared to healthy controls. MDD patients with SI had greater GMV of bilateral ACC compared to MDD patients without SI, and (3) GMV of bilateral ACC was positively correlated with SI in MDD patients and entered the regression equation in the subsequent logistic regression analysis. Thus, we found that there was a significant difference in GMV of ACC between MDD patients with and without SI, and that GMV of ACC was associated with SI in MDD patients.
Our study found that MDD patients with SI had higher HAMD score compared to MDD patients without SI, suggesting that MDD patients with SI may have more severe depressive symptoms, which is consistent with previous findings (Cai et al., Reference Cai, Jin, Liu, Zhang, Zhang, Cheung and Xiang2021; Guo et al., Reference Guo, Jiang, Jia, Zhu, Han, Wu and Tang2023; Khandoker et al., Reference Khandoker, Luthra, Abouallaban, Saha, Ahmed, Mostafa and Jelinek2017).
As the first step in suicidal behavior, SI is the product of the interaction of multiple risk factors, such as genetic and environmental factors (Dada et al., Reference Dada, Qian, Al-Chalabi, Kolla, Graff, Zai and De Luca2021; Ropaj, Reference Ropaj2023). Most psychiatric disorders increase suicide risk, especially depression (Cai et al., Reference Cai, Jin, Liu, Zhang, Zhang, Cheung and Xiang2021; Dada et al., Reference Dada, Qian, Al-Chalabi, Kolla, Graff, Zai and De Luca2021; Dai et al., Reference Dai, Zhou, Liu, Wei, Zhou, Tian and Zhang2022; Nassan et al., Reference Nassan, Daghlas, Winkelman, Dashti and Saxena2022; Schneider, Chen, Lungu, & Grasso, Reference Schneider, Chen, Lungu and Grasso2020). In patients with MDD, a higher prevalence of SI tends to be associated with more severe depressive symptoms (Cai et al., Reference Cai, Jin, Liu, Zhang, Zhang, Cheung and Xiang2021). A variety of treatments, including electroconvulsive therapy, repetitive transcranial magnetic stimulation (rTMS), and medications, can simultaneously alleviate patients’ depressive symptoms and reduce their SI (Hetrick et al., Reference Hetrick, McKenzie, Bailey, Sharma, Moller, Badcock and Meader2021; Hochschild et al., Reference Hochschild, Keilp, Madden, Burke, Mann and Grunebaum2022; Ionescu et al., Reference Ionescu, Fu, Qiu, Lane, Lim, Kasper and Canuso2021; Li, Yu, et al., Reference Li, Yu, Huang, Chen, Ai, Zhou and Kuang2021; Pan et al., Reference Pan, Shen, Jiao, Chen, Li, Lu and Huang2020). Serum cystatin C (Cys C) exerts biological functions in several aspects including bioactivity and neurophysiology, and it may affect the risk of depression in a number of ways (Daria et al., Reference Daria, Proma, Shahriar, Islam, Bhuiyan and Islam2020; Islam et al., Reference Islam, Ali, Karmoker, Kadir, Ahmed, Nahar and Islam2020; Zi & Xu, Reference Zi and Xu2018). Changes in Cys C levels may be associated with increased neuronal inflammation, which may increase some of the inflammatory factors involved in suicidal tendency in depressed patients (Sun, Chen, & Li, Reference Sun, Chen and Li2021). For example, IL-6 and TNF-α affect suicide risk in depressed patients by affecting the serotonergic system (Brundin, Bryleva, & Thirtamara Rajamani, Reference Brundin, Bryleva and Thirtamara Rajamani2017; Pandey et al., Reference Pandey, Rizavi, Ren, Fareed, Hoppensteadt, Roberts and Dwivedi2012). Cys C may also play an important role in suicide risk in depressed patients by inducing neuronal apoptosis and disrupting WM and brain amyloid deposition (Jia et al., Reference Jia, Huang, Wu, Zhang, Lui, Zhang and Gong2010; Mitaki et al., Reference Mitaki, Nagai, Sheikh, Terashima, Isomura, Nabika and Yamaguchi2011; Sun et al., Reference Sun, Chen and Li2021). Furthermore, traumatic experiences are often associated with more severe depression, which in turn may lead to more generalized SI (Cai et al., Reference Cai, Jin, Liu, Zhang, Zhang, Cheung and Xiang2021). Although depressive symptoms in MDD patients are closely associated with SI, they remain independent of each other. Individuals with SI have higher levels of anhedonia than those without SI, but depression does not explain the association between high levels of anhedonia and SI (Ducasse et al., Reference Ducasse, Loas, Dassa, Gramaglia, Zeppegno, Guillaume and Courtet2018). Therefore, the relationship between depressive symptoms and SI needs further research.
The present study demonstrated that the GMV in the ACC region was significantly smaller in MDD patients compared to healthy controls. Consistent with previous findings, MDD patients exhibited ACC atrophy compared to healthy controls (He et al., Reference He, Ping, Chu, Zeng, Shen and Xu2022). This atrophy may be associated with MDD, a degenerative mental disorder (Schmaal et al., Reference Schmaal, Pozzi, Tiffany, van Velzen, Veer, Opel and Veltman2020; Stein et al., Reference Stein, Meller, Brosch, Schmitt, Ringwald, Pfarr and Kircher2021; Zhang, Wei, et al., Reference Zhang, Wei, Chang, Jiang, Tang and Wang2020). The ACC is a key region associated with MDD (Cole et al., Reference Cole, Stimpson, Bentzley, Gulser, Cherian, Tischler and Williams2020; Crowell et al., Reference Crowell, Riva-Posse, Holtzheimer, Garlow, Kelley, Gross and Mayberg2019; Rappaport, Kandala, Luby, & Barch, Reference Rappaport, Kandala, Luby and Barch2020). In a study using rTMS to treat MDD, the left ACC of the pathological neural network was found to be connected to the DLPFC, and the abnormalities of the frontal and parietal lobes were improved after treatment (Belleau et al., Reference Belleau, Treadway and Pizzagalli2019; Philip et al., Reference Philip, Barredo, van t Wout-Frank, Tyrka, Price and Carpenter2018).
However, in this study, the GMV of the bilateral ACC was greater in MDD patients with SI compared to MDD patients without SI, suggesting that there is a significant difference in the GMV of the ACC between MDD patients with and without SI, as evidenced by the significantly lower atrophy of the bilateral ACC of MDD patients with SI. This is consistent with previous findings in MDD patients with higher rates of suicide attempts and type I bipolar disorder (Duarte et al., Reference Duarte, Neves, Albuquerque, Turecki, Ding, de Souza-Duran and Correa2017; Rizk et al., Reference Rizk, Rubin-Falcone, Lin, Keilp, Miller, Milak and Mann2019). The ACC plays an important role in cognitive reappraisal of negative emotions and serves the attentional control network (Buhle et al., Reference Buhle, Silvers, Wager, Lopez, Onyemekwu, Kober and Ochsner2014; Shenhav, Cohen, & Botvinick, Reference Shenhav, Cohen and Botvinick2016). Previous studies have shown that a subset of suicide attempters with higher lethality have higher levels of pre-suicide planning and lower levels of delay discounting, so that they can suppress their desire for immediate gratification and choose delayed rewards (Chaudhury et al., Reference Chaudhury, Singh, Burke, Stanley, Mann, Grunebaum and Oquendo2016; Dombrovski et al., Reference Dombrovski, Szanto, Siegle, Wallace, Forman, Sahakian and Clark2011). Suicide attempters with higher lethality also perform better on an object alternation task compared to those with lower lethality, suggesting greater executive control and response organization (Keilp et al., Reference Keilp, Wyatt, Gorlyn, Oquendo, Burke and John Mann2014). Reduced globus pallidus volume has also been found in suicide attempters with a strong desire to die and in those with higher lethality rates, which is associated with their non-impulsive temperament (Dombrovski et al., Reference Dombrovski, Siegle, Szanto, Clark, Reynolds and Aizenstein2012; Vang, Ryding, Traskman-Bendz, van Westen, & Lindstrom, Reference Vang, Ryding, Traskman-Bendz, van Westen and Lindstrom2010). MDD patients with SI have greater GMV in the ACC, which may be associated with their better cognitive control and overlap with suicide attempters, with higher mortality due to suicidal behaviors, representing a non-impulsive subgroup (Rizk et al., Reference Rizk, Rubin-Falcone, Lin, Keilp, Miller, Milak and Mann2019). At the same time, increased GMV may represent a compensatory process for the prolonged high stress prior to the onset of suicidal behavior (Rizk et al., Reference Rizk, Rubin-Falcone, Lin, Keilp, Miller, Milak and Mann2019). Increased microglial density in suicidal individuals suggests that microglial activation may be a result of pre-suicidal stress (Steiner et al., Reference Steiner, Bielau, Brisch, Danos, Ullrich, Mawrin and Bogerts2008). This microglia activation leads to increased somatic cell size and coarsening of branching processes, a change that may be reflected in the larger GMV observed in the current study (LaVoie, Card, & Hastings, Reference LaVoie, Card and Hastings2004; Rizk et al., Reference Rizk, Rubin-Falcone, Lin, Keilp, Miller, Milak and Mann2019). Finally, heterogeneity in suicidal behavior, especially in terms of lethality and intent, may contribute to inconsistent results (Rizk et al., Reference Rizk, Rubin-Falcone, Lin, Keilp, Miller, Milak and Mann2019). Suicidal behaviors are diverse, ranging from highly lethal suicide attempts to less lethal ones; from impulsive suicidal behaviors to well-planned and determined suicidal behaviors (Chaudhury et al., Reference Chaudhury, Singh, Burke, Stanley, Mann, Grunebaum and Oquendo2016; Keilp et al., Reference Keilp, Wyatt, Gorlyn, Oquendo, Burke and John Mann2014; Rizk et al., Reference Rizk, Rubin-Falcone, Lin, Keilp, Miller, Milak and Mann2019). Overrepresentation of one type of study may produce different results than other studies (Rizk et al., Reference Rizk, Rubin-Falcone, Lin, Keilp, Miller, Milak and Mann2019). Individuals with highly lethal suicide attempts may be similar to suicidal individuals in terms of demographics, clinical presentation, and brain biology (Chaudhury et al., Reference Chaudhury, Singh, Burke, Stanley, Mann, Grunebaum and Oquendo2016; Rizk et al., Reference Rizk, Rubin-Falcone, Lin, Keilp, Miller, Milak and Mann2019). Our results are similar to those of another study of MDD patients with a history of at least one highly lethal suicide attempt who were not receiving medication, suggesting that GMV is greater in patients with SI (Rizk et al., Reference Rizk, Rubin-Falcone, Lin, Keilp, Miller, Milak and Mann2019). However, the physiological mechanisms underlying increased GMV remain to be further investigated.
In this study, GMV in bilateral ACC was positively correlated with SI in MDD patients. Subsequent regression analysis showed that the GMV of cluster significantly associated with SI in MDD patients was related to SI in MDD patients. The ACC is a structure in the medial prefrontal cortex consisting of multiple functional divisions (Palomero-Gallagher et al., Reference Palomero-Gallagher, Hoffstaedter, Mohlberg, Eickhoff, Amunts and Zilles2019). The ACC is involved in emotion regulation, thought-emotion valence, decision-making, and impulsivity, and is central to neuroanatomical models of depression (Gradone et al., Reference Gradone, Champion, McGregor, Nocera, Barber, Krishnamurthy and Dotson2023; van Heeringen, Bijttebier, Desmyter, Vervaet, & Baeken, Reference van Heeringen, Bijttebier, Desmyter, Vervaet and Baeken2014; Wise et al., Reference Wise, Radua, Via, Cardoner, Abe, Adams and Arnone2017). There is accumulating evidence that ACC abnormalities are associated with suicidality across different diagnostic categories of psychiatric disorders, including MDD (He et al., Reference He, Ping, Chu, Zeng, Shen and Xu2022; Wagner et al., Reference Wagner, Koch, Schachtzabel, Schultz, Sauer and Schlosser2011), schizophrenia (van Heeringen et al., Reference van Heeringen, Bijttebier, Desmyter, Vervaet and Baeken2014), and borderline personality disorder (Duarte et al., Reference Duarte, Neves, Albuquerque, Turecki, Ding, de Souza-Duran and Correa2017). This transdiagnostic perspective is supported by recent studies that show the consistency in certain brain networks despite current psychopathology (Goodkind et al., Reference Goodkind, Eickhoff, Oathes, Jiang, Chang, Jones-Hagata and Etkin2015; Gradone et al., Reference Gradone, Champion, McGregor, Nocera, Barber, Krishnamurthy and Dotson2023; Wise et al., Reference Wise, Radua, Via, Cardoner, Abe, Adams and Arnone2017). In this study, SI increased with increasing GMV in the ACC of MDD patients. One possible explanation is that overactivity of the ACC may lead to increased brain plasticity and thus become larger (Fears et al., Reference Fears, Schur, Sjouwerman, Service, Araya, Araya and Bearden2015; Lisy et al., Reference Lisy, Jarvis, DelBello, Mills, Weber, Fleck and Adler2011). This could act as a compensatory mechanism for regulating emotional states, mitigating failures in frontal lobe “top-down” regulation (Duarte et al., Reference Duarte, Neves, Albuquerque, Turecki, Ding, de Souza-Duran and Correa2017). Furthermore, epigenetic dysregulation of glucocorticoid receptors and serotonin receptor binding plays an important role in this process (Bartlett et al., Reference Bartlett, Zanderigo, Shieh, Miller, Hurley, Rubin-Falcone and Mann2022; Pantazatos et al., Reference Pantazatos, Melhem, Brent, Zanderigo, Bartlett, Lesanpezeshki and Mann2022; Rizavi et al., Reference Rizavi, Khan, Zhang, Bhaumik, Grayson and Pandey2023). Taken together, ACC may be involved in the pathological mechanism of SI in MDD patients, suggesting that GMV in the ACC may be a biomarker for SI in sensitive MDD populations.
In contrast to our findings, previous studies have shown that the GMV of ACC is smaller in MDD patients with suicide attempts compared to MDD patients without suicide attempts and healthy controls, and that the GMV of ACC progressively decreases with increasing SI in MDD patients (He et al., Reference He, Ping, Chu, Zeng, Shen and Xu2022; Wagner et al., Reference Wagner, Koch, Schachtzabel, Schultz, Sauer and Schlosser2011). Differences from the current study may be due to heterogeneity in the degree of suicide attempts and suicide fatalities, as well as differences in the way SI subgroups were distinguished (Klonsky, May, & Saffer, Reference Klonsky, May and Saffer2016). Similarly, a previous study used only 30 individuals recruited from the inpatient and outpatient services of the Department of Psychiatry and Psychotherapy at Friedrich-Schiller University (Wagner et al., Reference Wagner, Koch, Schachtzabel, Schultz, Sauer and Schlosser2011). Only 129 MDD patients were used in another study with Han Chinese (He et al., Reference He, Ping, Chu, Zeng, Shen and Xu2022). It is different from the large sample size of Chinese Han individuals in this study. In addition, the use of different methods such as Freesurfer and VBM in earlier versions of the data processing software SPM2 may have led to differences in the results (He et al., Reference He, Ping, Chu, Zeng, Shen and Xu2022; Wagner et al., Reference Wagner, Koch, Schachtzabel, Schultz, Sauer and Schlosser2011).
There is evidence that a long disease duration is associated with a poor prognosis in MDD patients (Altamura, Serati, & Buoli, Reference Altamura, Serati and Buoli2015). Long disease duration impairs the immune system and central nervous system (Altamura et al., Reference Altamura, Serati and Buoli2015; Arango, Breier, McMahon, Carpenter, & Buchanan, Reference Arango, Breier, McMahon, Carpenter and Buchanan2003). This process is associated with glutamate-mediated neuronal cytotoxicity and activation of the hypothalamic–pituitary–adrenal (HPA) axis (Matsuo et al., Reference Matsuo, Izumi, Kume, Takada-Takatori, Sawada and Akaike2010). This biological abnormality may be the result of a neurodegenerative model (Altamura, Buoli, & Pozzoli, Reference Altamura, Buoli and Pozzoli2014; Skeppar et al., Reference Skeppar, Thoor, Agren, Isakson, Skeppar, Persson and Fitzgerald2013). Atypical antipsychotics and antidepressants with manic and mood-stabilizing effects can block the neurodegenerative process associated with long disease duration, especially during the first year of illness in patients with first episode MDD (Andreasen, Reference Andreasen2010; Ho, Andreasen, Ziebell, Pierson, & Magnotta, Reference Ho, Andreasen, Ziebell, Pierson and Magnotta2011; van Haren et al., Reference van Haren, Schnack, Cahn, van den Heuvel, Lepage, Collins and Kahn2011). Long disease duration and low medication adherence may contribute to an increased risk of SI in patients with MDD (Altamura et al., Reference Altamura, Serati and Buoli2015). Antidepressants are first-line medications for the treatment of severe MDD patients and have been shown to improve dysfunction and alter neural activation and brain structure (Cleare et al., Reference Cleare, Pariante, Young, Anderson, Christmas and Cowen2015; Dichter, Gibbs, & Smoski, Reference Dichter, Gibbs and Smoski2015). MRI of MDD patients taking antidepressants has shown changes in GMV (Dusi, Barlati, Vita, & Brambilla, Reference Dusi, Barlati, Vita and Brambilla2015). Remission of depressive symptoms after medication is associated with increased GMV in the right ACC, increased functional and metabolic activity, and altered connectivity in the cingulate bundle (Korgaonkar, Williams, Song, Usherwood, & Grieve, Reference Korgaonkar, Williams, Song, Usherwood and Grieve2014). The GMV of the bilateral ACC may be particularly sensitive to the clinical response to pharmacological treatment (Liu et al., Reference Liu, Xu, Luo, Luo, Chen, Lui and Gong2017). In this study, the results were still statistically significant when first episode, medication status, and disease duration were included as covariates. Therefore, the results of the current study are robust.
Limitations
It is worth noting that this study has several limitations. First, participants in this study were recruited from multiple clinical centers, and therefore some of the clinical data were incomplete, such as the name of the medication used, the dosage, and the parameters of the MRI scanner. Second, SI was collected through the HAMD item 3 rather than a structured SI-specific instrument. Replication based on structured SI-specific instruments is needed for this study. Third, the lack of raw data did not allow us to further analyze functional indicators, such as functional brain connectivity in the ACC region in MDD patients with and without SI. Fourth, the database did not specifically record whether each patient was diagnosed through DSM-IV or ICD-10. Therefore, it is unclear how many enrolled MDD patients were diagnosed with ICD-10 and DSM-4, respectively. This gap should be remedied in future studies by using the same criteria or by precise documentation. Finally, there was a large difference in HAMD scores between MDD patients with and without SI, thus further studies are needed to investigate the relationship between depression severity and SI and structural brain abnormalities.
Conclusion
In conclusion, our study found that the GMV in the ACC region was significantly smaller in MDD patients compared to healthy controls. Compared to MDD patients without SI, MDD patients with SI had higher HAMD scores, greater GMV of the bilateral ACC, and significantly less atrophy. The GMV in the bilateral ACC of MDD patients was positively correlated with SI. We hypothesized that ACC may be involved in the pathological mechanism of SI in MDD patients, and GMV of ACC may be a biomarker of SI in MDD patients. However, there were methodological limitations of this study, including incomplete clinical data, no use of a specific structured SI tool, lack of raw data, and differences in HAMD scores between MDD patients with and without SI. Therefore, there is a need to assess MDD patients with small differences in HAMD scores using a specific structured SI tool, and to collect their complete clinical data and raw data to understand the mechanism of differences in GMV of ACC between MDD patients with and without SI.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0033291724003283.
Data availability statement
The data that support the findings of this study are available from the corresponding author, (ZXY), upon reasonable request.
Acknowledgments
Members of the REST-meta-MDD Consortium provided the data for this study. The authors would like to express their appreciation and gratitude to all the participants in this study.
Author contribution
LX and XYZ performed analysis, made interpretation of data, and drafted the manuscript. YCG made contributions to the acquisition of clinical and MRI data for the work. NW, DW, and HZ helped draft the work and revise it critically. All authors made significant contributions to the paper to assess the important intellectual content, read, and approved the paper.
Financial support
This study supported by the Academic Research Projects of Beijing Union University (ZK10202301). The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Competing interest
No conflict of interest was disclosed for each author.
Ethical standard
The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.
Open access
