Neuropathological characteristics of abnormal white matter functional signaling in adolescents with major depression

Abstract

BACKGROUND

Major depression disorder (MDD) constitutes a significant mental health concern. Epidemiological surveys indicate that the lifetime prevalence of depression in adolescents is much higher than that in adults, with a corresponding increased risk of suicide. In studying brain dysfunction associated with MDD in adole-scents, research on brain white matter (WM) is sparse. Some researchers even mistakenly regard the signals generated by the WM as noise points. In fact, studies have shown that WM exhibits similar blood oxygen level-dependent signal fluctuations. The alterations in WM signals and their relationship with disease severity in adolescents with MDD remain unclear.

AIM

To explore potential abnormalities in WM functional signals in adolescents with MDD.

METHODS

This study involved 48 adolescent patients with MDD and 31 healthy controls (HC). All participants were assessed using the Patient Health Questionnaire-9 Scale and the mini international neuropsychiatric interview (MINI) suicide inventory. In addition, a Siemens Skyra 3.0T magnetic resonance scanner was used to obtain the subjects' image data. The DPABI software was utilized to calculate the WM signal of the fractional amplitude of low frequency fluctuations (fALFF) and regional homogeneity, followed by a two-sample t-test between the MDD and HC groups. Independent component analysis (ICA) was also used to evaluate the WM functional signal. Pearson’s correlation was performed to assess the relationship between statistical test results and clinical scales.

RESULTS

Compared to HC, individuals with MDD demonstrated a decrease in the fALFF of WM in the corpus callosum body, left posterior limb of the internal capsule, right superior corona radiata, and bilateral posterior corona radiata [P < 0.001, family-wise error (FWE) voxel correction]. The regional homogeneity of WM increased in the right posterior limb of internal capsule and left superior corona radiata, and decreased in the left superior longitudinal fasciculus (P < 0.001, FWE voxel correction). The ICA results of WM overlapped with those of regional homo-geneity. The fALFF of WM signal in the left posterior limb of the internal capsule was negatively correlated with the MINI suicide scale (P = 0.026, r = -0.32), and the right posterior corona radiata was also negatively correlated with the MINI suicide scale (P = 0.047, r = -0.288).

CONCLUSION

Adolescents with MDD involves changes in WM functional signals, and these differences in brain regions may increase the risk of suicide.

Key Words: White matter, Regional homogeneity, The fractional amplitude of low-frequency fluctuations, Independent component analysis, Adolescents, Major depression disorders

Core Tip: This groundbreaking study investigates white matter (WM) functional signals in adolescents with major depressive disorder (MDD), an area often overlooked in research. Utilizing advanced imaging techniques, the study identifies specific abnormalities in WM signals, revealing decreased fractional amplitude of low frequency fluctuations in key regions and altered regional homogeneity and independent component analysis patterns. Notably, these changes correlate with suicidality scales, indicating a potential link between WM anomalies and severity of depression. The study pioneers a crucial shift in understanding MDD's neuropathogenesis, offering novel insights and support for future research and predictive measures.

INTRODUCTION

Major depressive disorder (MDD) is a widespread psychiatric condition across the globe, with an estimated lifetime prevalence rate of approximately 11% among adolescents[1,2]. It has been reported that over half of adolescent suicide victims had a depressive disorder[3]. MDD in adolescents is associated with an increased risk of suicide, and the disorder often persists into adulthood[4]. Therefore, the increasing incidence of depression among adolescents requires urgent attention[5]. The recognition and treatment of MDD in adolescence are crucial; however, our current grasp of the physiological and pathological underpinnings of the condition remains incomplete.

Functional magnetic resonance imaging (fMRI) is a non-invasive technique, and it can be used to indirectly measure neuronal activity via the blood oxygenation level-dependent (BOLD) signal[6]. In recent years, the development and progress of magnetic resonance imaging (MRI) technology have provided an opportunity to study the pathophysiology of MDD using various resting state fMRI (rs-fMRI) techniques. These techniques have been extensively utilized in the exploration of the physiological and pathological foundations of the brain in mental disorders. The changes in the resting state of gray matter (GM) found in most studies are often used as biomarkers for neuropsychiatric disorders[7-9], while white matter (WM) tends to be ignored[10,11], even though the volume of WM in the human brain accounts for approximately 40%-45%[12].

To date, various approaches are used to analyze spontaneous BOLD signals, such as the amplitude of low frequency fluctuations (ALFF), the fractional ALFF (fALFF), regional homogeneity (ReHo), and independent component analysis (ICA)[13]. The fALFF[14] is a metric believed to capture spontaneous neural activity and has been demonstrated to correlate with regional brain glucose metabolism[15-17], which effectively mitigates physiological noise when compared to ALFF[14]. ReHo, a data-driven approach, implies that the time series of spatially adjacent voxels exhibit greater temporal similarity when the brain region is engaged in a specific condition[18]. ICA is also a data-driven method, it can decompose fMRI data into spatially independent and functionally connected brain networks[19]. ReHo and ICA can acquire a greater amount of information than methods driven by models[18,20]. Using these data analysis methods, Liu et al[21] discovered the alterations in ReHo and ALFF in the precentral gyrus, postcentral gyrus, and paracentral gyrus in MDD[21]. In addition, a meta-analysis found a correlation between amygdala activity and depression[22]. These findings may indicate the pathological and physiological processes associated with MDD.

Recent studies have shown the presence of functional brain activity related to neuronal activity in the WM, including connectivity and interconnection functions[23,24]. In task related studies, activation of brain regions can be detected in the inner capsule and corpus callosum (CC)[25,26]. Additional studies showed that fMRI activity within particular WM pathways is remarkably consistent during the resting state, and indicated that these WM signals exhibit features remi-niscent of hemodynamic (BOLD) alterations linked to neuronal activity[27,28]. Furthermore, a recent study employing ICA and hierarchical clustering revealed the presence of clusters of correlated activity within the WM[29]. These findings reveal that WM may play a crucial role in resting states. We hope to provide more information in order to understand the underlying pathological mechanisms of adolescent severe depression by revealing the fluctuation characteristics of WM functional signals.

Therefore, we decided to use the GM analysis method to explore changes in WM, using a combination of multiple features, including fALFF, ReHo, and ICA. These three analytical methods present a progressive relationship layer by layer, from individual, to local, and finally to component networks. The combination of multiple analytical methods is used to explore the functional differences of WM in adolescents with MDD. This method is expected to reveal the characteristics and potential biological mechanisms of abnormal brain activity in adolescents with MDD during the resting state, filling the current incomplete understanding of the physiological and pathological basis of the disease. From this study, we may not only better understand the pathogenesis of MDD in adolescents, but also provide more targeted methods for future diagnosis and treatment, thereby more effectively addressing this global health problem.

MATERIALS AND METHODS

Participants

A total of 84 subjects, including 51 adolescents with MDD and 33 healthy controls (HC) matched by age, gender, education and right-handedness were initially recruited from Suzhou Guangji Hospital. Using the mini international neuropsychiatric interview (MINI), patients were diagnosed by two trained psychiatrists above the attending level who conducted a structured interview. The inclusion criteria were: (1) The patients had not received systematic medication treatment prior to the MRI scan; (2) Patient Health Questionnaire-9 (PHQ-9) scores ≥ 20; (3) Right-handed; (4) Met the criteria of the Diagnostic and Statistical Manual of Mental Disorders, 5^th edition; and (5) Aged 11 to 18 years. The exclusion criteria were: (1) Contrain-dications to MRI; (2) Any other psychiatric disorders such as bipolar disorder and schizophrenia; (3) Individuals with organic brain diseases, as identified by imaging; and (4) 32-item Hypomania Checklist (HCL-32) scores < 14.

This study was approved by the Ethics Committee of Suzhou Guangji Hospital, and each subject signed a written informed consent form. For participants under age 18, at least one legal guardian signed an informed consent form on their behalf.

Clinical assessments

Two attending psychiatrists with specialized training assessed the clinical symptoms of the participants and performed a reliability assessment. The PHQ-9 is a reliable and valid screening tool for depression[30,31], derived from the depression section of the Patient Health Questionnaire developed by Spitzer et al[32]. All participants completed the MINI suicidality subscale (MINISS), a user-friendly and highly accurate tool for predicting suicide risk[33] and the HCL-32, a self-assessment tool for hypomanic symptoms[34].

MRI acquisition

All participants’ image data were obtained on a 3.0T Siemens Skyra scanner at Suzhou Guangji Hospital, equipped with a head/neck 20 channel coil. The scanning parameters of rs-fMRI are as follows: Repetition time = 2000 ms, echo time = 30 ms, slice thickness = 3.5 mm, 32 slices, slice gap = 0.875 mm, filed of view = 224 mm × 224 mm, flip angle = 90°, matrix size = 64 voxels × 64 voxels, acquisition time = 8.04 min. During the scan procedure, each subject was asked to lie flat in the machine, and close their eyes but not to fall asleep. While the subjects were being scanned, a sponge pad was placed on each person's head to prevent head movement and obtain clear images.

Data preprocessing

The MRI data of each subject were preprocessed by The Data Processing & Analysis for (Resting-state) Brain Imaging (DPABI)[35] and Statistical Parametric Map-ping (SPM12, http://www.fil.ion.ucl.ac.uk/spm) toolkit in MATLAB 2016b. This was based on the following steps: (1) Conversion data format from DICOM to NIFTI; (2) the first 10 time points were removed in order to stabilize the data; (3) slice timing and realignment of head motion correction (any participant whose head motion exceeded 2.0 mm or rotation exceeded 2.0° was excluded); (4) the T1 images were registered to functional images and segmentation into WM and GM and cerebrospinal fluid using the New Segment algorithm; (5) apply white mask to functional images; (6) normalize the functional image space to a standard space (Montreal Neurological Institute) using DARTEL, and resampled to a voxel size of 3 mm × 3 mm × 3 mm; and (7) extraction of individual-level WM 4D images[36]. Five participants were excluded due to head motion > 2 mm or 2°. Finally, 48 MDD and 31 HC were included for further analysis in the current study.

fALFF and ReHo of WM calculation

Using a fast Fourier transform at each voxel, we computed the power of the BOLD signal of WM within the low-frequency range of 0.01–0.10 Hz and subsequently divided it by the entire frequency range to calculate the fALFF of WM[14].

The calculation of ReHo of WM values was performed as follows: Firstly, a low-pass filter (0.01–0.1 Hz) was used to mitigate the effects of high-frequency noise and low-frequency drift. Then, Kendall's Coefficient of Concordance[18], also known as the ReHo value, was utilized to assess the similarity between an individual voxel and its neighboring 27 voxels. Subsequently, in order to minimize individual variance, the ReHo value for each voxel was normalized by dividing it by the global mean ReHo value.

Extraction of WM-fMRI signals via ICA

On the basis of the aforementioned WM signal preprocessing, the preprocessed imaging data in each group was used to perform group ICA analysis using the fMRI toolbox (GIFT, version 3.0C, http://mialab.mrn.org/software/gift)[37]. Firstly, the pre-processed data's dimensionality was reduced, which was followed by application of the Infomax algorithm for spatial ICA on this data. In addition, component stability was attained by running 100 iterations in a software package called ICASSO[38]. A total of 6 components were estimated. Finally, we selected an interesting independent component that was related to the alteration of WM in MDD for further evaluation.

Statistical analysis

Differences in demographics between the two participant groups were assessed using SPSS version 29.0, sex differences were assessed using the chi-squared test, while other parameters were compared between the two groups using a two-sample t-test.

Voxel-based comparisons of fALFF and ReHo of WM maps between the patient and control groups were performed using the two-sample t-test in DPABI software with age, sex, education, head-motion included as covariates, and with a threshold setting at P < 0.001, FWE voxel correction for multiple comparisons. Additionally, sex, age, education, and head motion were controlled for during the analysis of ICA maps. The selected component of interest was then compared between the patient and control groups (FWE voxel correction for multiple comparisons).

Correlational analyses were performed to investigate the association between fALFF, ReHo values of MW in regions exhibiting group differences and clinical variables, including the PHQ-9 and the MINI Suicide Inventory.

RESULTS

Demographic and clinical variables

The demographic and clinical characteristics of the adolescents with MDD and HC are summarized in Table 1. There were no significant differences in sex, age, education and head-motion between the two groups.

Table 1 Demographic and clinical characteristics of the included subjects.

	MDD (n = 48)	HC (n = 31)	χ²/t	P value
Sex (male/female)	8/40	6/25	0.093	0.771
Age (yr)	14.15 (1.79)	14.71 (2.00)	-1.306	0.195
Education (yr)	8.44 (1.73)	8.87 (1.77)	-1.081	0.283
Head-motion	0.08 (0.04)	0.07 (0.03)	0.234	0.815
PHQ-9 scores	23.02 (2.39)	-	-	-
HCL-32 scores	6.94 (2.77)	-	-	-
MINISS scores	22.44 (9.61)	-	-	-

PHQ-9: Patient Health Questionnaire-9; HCL-32: The 32-item Hypomania Checklist; MINISS: The mini international neuropsychiatric interview suicidality subscale; MDD: Major depression disorders; HC: Healthy controls.

fALFF and ReHo of WM alterations in adolescents with MDD

WM brain regions that exhibited disparities between the groups in the fALFF and ReHo analyses were identified and reported using the JHU-ICBM WM label atlas (provided by Wakana et al[39] and Hua et al[40] from the Laboratory of Brain Anatomical MRI at Johns Hopkins University)[39,40]. With regard to the fALFF calculation, patients with MDD showed decreased fALFF in the left posterior limb of the internal capsule (PLIC), posterior corona radiata (PCR), right PCR, superior corona radiata (SCR), and CC body (P < 0.001, FWE voxel correction for multiple comparisons). With regard to the ReHo computation, MDD patients displayed decreased ReHo values in the left superior longitudinal fasciculus (SLF), and higher ReHo values in the right PLIC and the left SCR (P < 0.001, FWE voxel correction for multiple comparisons). These results are presented in Table 2 and Figure 1.

Figure 1 White matter regions with significant differences in fractional amplitude of low frequency fluctuations, regional homogeneity and independent component analysis in adolescents with major depression disorders compared to healthy controls. Maps a threshold at P < 0.001, FWE voxel correction. fALFF: Fractional amplitude of low frequency fluctuations; WM: White matter; ReHo: Regional homogeneity; ICA: Independent component analysis; L: Left; R: Right; MDD: Major depression disorders.

Table 2 Brain regions with significant differences in fractional amplitude of low frequency fluctuations, regional homogeneity and independent component analysis between adolescents with major depression disorders and healthy controls.

Indices	Anatomical region	MNI coordinates, x, y, z	Peak intensity	Cluster size
fALFF of WM	Posterior limb of the internal capsule L	21, -18, 3	-6.12	27
	Posterior corona radiata R	-24, -39, 33	-5.57	36
	Body of the corpus callosum	12, 15, 30	-5.75	62
	Superior corona radiata R	-18, -21, 51	-5.44	71
	Posterior corona radiata L	24, -36, 39	-6.39	106
ReHo of WM	Posterior limb of the internal capsule R	27, 0, 18	5.40	60
	Superior corona radiata L	-27, 3, 39	4.55	23
	Superior longitudinal fasciculus L	-30, -39, 42	-5.33	18
	Superior longitudinal fasciculus L	-24, -24, 48	-4.76	14
ICA of WM	Posterior limb of the internal capsule R	27, -15, 12	4.30	26
	Superior longitudinal fasciculus L	-24, -24, 45	-5.80	25

The significance threshold was set at P < 0.001, FWE voxel correction for multiple comparisons. fALFF: Fractional amplitude of low frequency fluctuations; WM: White matter; ReHo: Regional homogeneity; ICA: Independent component analysis; L: Left; R: Right; MNI: Montreal neurological institute.

Spatial ICA of rs-fMRI signals of the WM

We investigated the spatiotemporal patterns in rs-WM-fMRI data using ICA. Six spatially independent components were estimated and extracted from the time series of all WM voxels. Subsequent analyses of the 6 components showed decreased connectivity of the left SLF and higher connectivity of the right PLIC in MDD patients relative to HC (P < 0.001, FWE voxel correction for multiple comparisons; Table 2 and Figure 1).

Correlation analysis

The fALFF and ReHo values were extracted from WM regions that displayed significant differences between adolescents with MDD and the HC group, and correlation analyses between these region’s values and clinical characteristics were conducted. We found there is no correlation between significant differential brain regions and PHQ-9 scales. A negative correlation between the fALFF values in the left PLIC and the MINI suicide scale (P = 0.026, r = -0.320; Figure 2A), as well as between the right PCR and the MINI suicide scale (P = 0.047, r = -0.288; Figure 2B).

Figure 2 Correlations between clinical psychiatric symptoms and white matter regions with significant differences in the major depression disorders and healthy controls groups. A: It is a significant brain region left posterior limb of internal capsule in fractional amplitude of low frequency fluctuations (fALFF) of white matter (WM); B: It is a significant brain region right posterior corona radiata in fALFF of WM. MINISS: The mini international neuropsychiatric interview suicidality subscale.

DISCUSSION

In this study, we demonstrated functional changes of WM by employing several distinct rs-fMRI techniques (ReHo, fALFF, and ICA), for the first time, on datasets from healthy subjects, and MDD patients. We found that adolescent patients with MDD showed significant differences in the CC body, left SLF, bilateral PLIC, PCR and SCR compared with the HC group. We also investigated the relationship between functional changes in regions of WM and the clinical features in patients with MDD. These results indicated that resting state functional metrics of WM can be valuable in investigating the pathophysiologic basis of MDD.

In the fALFF results, significant differences in MDD patients were mainly observed in the CC, PLIC, and corona radiata, when compared with the HC group. The CC is a dense bundle of nerve fibers that plays a crucial role in connecting different regions of the neocortex. It facilitates neural circuits involved in cognitive and emotional processing[41,42]. In addition, the CC body contains fibers connecting the cingulate cortex, insular cortex, and temporal cortex[43], these areas are often associated with depression[44-46]. A lower fALFF value indicates a decrease in WM integrity of the CC, which may hinder the interaction between the cerebral hemispheres and cause emotional processing disorders in depression. It is worth noting that in this study, we found that adolescent patients with severe depression experienced abnormal activation of brain regions in the posterior limbs of the inner capsule using various analytical methods. This may be because the core symptom of MDD is low mood, and the PLIC participates in the formation of a neural network by connecting structures such as the cerebral cortex and hypothalamus. Its subcortical area belongs to the frontal striatal circuit[47], and these two circuits play a crucial role in emotional, cognitive, and motor functions[48]. Interestingly, in this study, MDD patients showed significantly lower fALFF values in the PLIC. In addition, Sisti et al[49] also found that cognitive decline was not only related to local brain lesions, but may also be related to the destruction of WM fibers and impaired connectivity in these brain regions[49]. Therefore, this discovery may explain the cognitive style of adolescent patients with depression. The corona radiata is composed of ascending and descending fibers that transmit information to the cerebral cortex and functionally involve emotions and executive processing[50,51]. Furthermore, in the correlation analysis, there was a negative correlation between the fALFF values of the PLIC and the PCR and the suicide scale. Some studies have found that impaired executive function may be a risk factor for suicide[52]. This may indicate that as brain dysfunction increases, the risk of suicide also increases.

The upper longitudinal bundle is considered the largest associative fiber bundle system in the brain[53], connecting the frontal and parietal lobes[54]. It is considered a higher-order multi-sensory associative system and is often reported to be related to executive function and emotions[55]. Previous studies have shown that the degree of damage to the SLF in patients with anxiety related depression may be more severe than in patients with non-anxiety related depression, which may lead to cognitive and emotional impairment[56]. This is consistent with the findings of this study in relation to WM ReHo. In a study of mild cognitive impairment (MCI), the left upper corona showed a lower fractional anisotropy value, suggesting that changes in WM in this brain area may be a potential biomarker of MCI[57]. Therefore, an abnormality of the SLF may indicate that patients with depression have more severe depression.

In this study, the ICA method was used to calculate the differences in the brain network of the components of interest. The brain regions overlapped with ReHo, and the direction of changes in the signal values of the brain regions was consistent. This may be because ReHo measures the local connectivity of spontaneous fMRI signals[18,58], and ICA studies measure inter-regional connectivity. These two methods are complementary to each other in a sense, which is why there is an abnormal overlap of activated brain regions. MDD patients also showed an increase in ReHo values in the left WM and a decrease in fALFF values on the right side in bilateral corona radiata lesions in this study, which may be due to the non-flow coupling metabolism of fALFF and ReHo.

Correlation analysis showed that significant correlations were observed between fALFF values, ReHo values, and clinical features in several WM regions. The left PLIC and the right PCR were negatively correlated with suicide. Research found that the bilateral PCR was associated with cognitive impairment in several different diseases[59-61]. There are many studies on the relationship between cognitive impairment and suicidal behavior[62,63]. In the study of suicidal ideation in schizophrenia, it is mentioned that the PCR may be associated with biological processes leading to depression and increased suicidal ideation[64]. This is consistent with the correlation between the PCR and the MINISS found in our study. In addition, studies have shown that impaired executive function may be a risk factor for suicide[52]. Therefore, we speculate that the abnormal activity of the right PCR and the left posterior limb brain area of the inner capsule may be potential biomarkers that trigger suicidal ideation in patients. There are multiple neural circuits within the inner capsule, and the corona radiata and fiber bundles within the capsule project from the cortex to the thalamus and pons nuclei[65]. The thalamus plays an important role in emotional regulation[66], and changes in the inner capsule may interfere with the connection between the thalamus and cortex, leading to abnormal emotional regulation and increasing the occurrence of manic symptoms.

There are some limitations in this study. Firstly, this was a cross-sectional study that failed to reveal the dynamic changes in WM functional signals over time in adolescents with severe depression. Further longitudinal research will help us understand the principles of this disease. Secondly, the sample size in this study was relatively small, and further research and verification are needed in a larger sample size. In addition, studies have also indicated that the relationship between BOLD signals observed in WM and neuronal related activities is still unclear[67]. We require more evidence in future work to demonstrate the importance of BOLD signals observed in WM.

CONCLUSION

Our research findings suggest that changes in WM functional signals may provide new insights into the neuro-physiological mechanisms of severe depression in adolescents, and that changes in WM functional signals may serve as biomarkers for predicting future trends in suicide in this disease.

ARTICLE HIGHLIGHTS

Research background

White matter (WM) is composed of various functional nerve fibers and plays an indispensable role in the central nervous system. However, the WM signal changes and their correlation with major depression disorder (MDD) in adolescents are still unclear.

Research motivation

An increasing number of studies have confirmed the functional organization of WM by the resting state functional magnetic imaging (rs-fMRI), indicating its feasibility of studying WM function in adolescents with MDD.

Research objectives

The purpose of this study is to explore the functional changes in the WM of adolescents with MDD.

Research methods

We collected rs-fMRI data and clinical scale information from the adolescent group with MDD and the healthy control group, and analyzed the correlation between WM function signals and clinical scales in the two groups.

Research results

We found significant changes in the functional signals of WM in adolescents with MDD, using the fractional amplitude of low frequency fluctuations, regional homogeneity, and independent component analysis. There are two brain regions, the left posterior limb of the inner capsule and the right posterior corona radiata, which are negatively correlated with the mini international neuropsychiatric interview suicide scales.

Research conclusions

The discovery of changes in WM functional signals in adolescents with MDD is of great significance for understanding the neuropathogenesis of depression.

Research perspectives

Our research findings may serve as biomarkers for predicting the risk of MDD and suicide in adolescents.

ACKNOWLEDGEMENTS

The corresponding author would like to thank all participants and all authors in the study for guidance and support.