Monitoring white matter volume alterations via cranial magnetic resonance imaging in major depressive disorder: Association with cognitive dysfunction

Abstract

BACKGROUND

Depression is considered a functional disorder, and routine brain magnetic resonance imaging (MRI) typically reveals no abnormalities. However, advances in diagnostic techniques and MRI modalities show that depression is associated with structural alterations in specific brain regions.

AIM

To examine the correlation between MRI-measured white matter volume changes and the severity of cognitive dysfunction in major depressive disorder (MDD).

METHODS

A retrospective review (January 2020 to June 2024) compared MDD patients (observation group) with healthy controls. Depression severity [17-item Hamilton Depression Rating Scale (HAMD-17)] and neuropsychological performance were assessed. Routine MRI and diffusion kurtosis imaging quantified fractional anisotropy (FA), mean kurtosis (MK), and mean diffusivity (MD) across bilateral frontal, temporal, and occipital white matter, and the corpus callosum genu/splenium. Correlations between these metrics and cognitive scores were evaluated.

RESULTS

Relative to controls, patients with MDD had higher HAMD-17 scores, lower Montreal Cognitive Assessment (MoCA) scores, longer completion times on the Trail Making Test Parts A and B (TMT-A and TMT-B), and reduced digit-symbol substitution performance (P < 0.05). FA was lower in the bilateral temporal lobes of the observation group (P < 0.05). MK values in the bilateral superior frontal gyrus and temporal lobes were reduced in patients with MDD (P < 0.05), while remaining similar in other regions (P > 0.05). MD values did not differ between groups (P > 0.05). In MDD, FA showed no significant associations with MoCA, TMT-A, TMT-B, or Digit Symbol Substitution Test (DSST) (P > 0.05). MK in the bilateral superior frontal gyrus correlated positively with MoCA (P < 0.05), and DSST scores were inversely associated with left superior frontal MK (P < 0.05).

CONCLUSION

Brain MRI can detect subtle structural abnormalities in the brain of patients with MDD. White matter fiber bundles in the bilateral superior frontal gyri and temporal lobes may exhibit atrophy, and these abnormalities show meaningful associations with cognitive dysfunction.

Key Words: Magnetic resonance imaging; Depression; Major depressive disorder; Diffusion kurtosis imaging; Fractional anisotropy; Mean kurtosis

Core Tip: Recent studies highlight notable neurobiological abnormalities in individuals with depression, with white matter lesions being relatively common. Magnetic resonance imaging (MRI), a non-invasive imaging method, clearly demonstrates the extent and distribution of such lesions. However, the causal relationship between depression and white matter lesions remains unclear, partly due to the complex interactions between depressive symptoms and cognitive impairment. This study quantified white matter volume changes using MRI and examined their correlation with cognitive impairment severity in major depressive disorder.

INTRODUCTION

Depression is a common mental illness and a major emotional disorder affecting hundreds of millions worldwide[1]. Characterized by persistent low mood, it is usually accompanied by cognitive disorders, psychomotor slowing, and emotional distress. Individuals with major depressive disorder (MDD) may experience marked hopelessness and, in severe cases, increased risk of self-harm, underscoring the need for timely intervention[2]. The recurrence rate of depression is reported to be 75%-80%, and repeated episodes can worsen symptoms and contribute to significant mortality, posing a serious challenge for families and society[3]. Growing evidence shows that depression is associated with significant neurobiological abnormalities and cognitive dysfunction. Cognitive deficits are typically assessed through neuropsychological testing and brain imaging, and advances in imaging technology have enabled more detailed investigations of brain structure[4]. Previous studies have identified white matter lesions as a relatively common pathological feature in depression[5]. However, the causal relationship between depression and white matter lesions remains unclear, partly due to the complex interactions between depressive symptoms and cognitive impairment.

Research on structural brain abnormalities in MDD began with computed tomography (CT)[6]. Because CT primarily reveals ventricular atrophy and demyelination in the semi-oval region, it can be difficult to distinguish these findings from normal aging. In recent years, magnetic resonance imaging (MRI) studies of depression have expanded, introducing new perspectives and hypotheses. MRI, a non-invasive neuroimaging method, can clearly delineate the extent and severity of white matter lesions[7]. Abnormalities in intercortical and cortico-subcortical neural circuits have been identified as key pathophysiological mechanisms of depression[8], and white matter lesions appear to relate with cognitive dysfunction[9]. As an emerging imaging modality in depression diagnosis, neuroimaging contributes objective evidence for diagnosis and prognosis. Diffusion kurtosis imaging (DKI) has advanced the evaluation of healthy populations, psychiatric disorders, geriatric diseases, and brain tumors[10,11]. By quantifying diffusion anisotropy and reconstructing white matter fiber pathways, DKI more accurately reflects the connectivity across brain regions and highlights microstructural differences not visible on routine imaging[12]. Therefore, this study retrospectively analyzes MDD cases, evaluates white matter microstructural changes using MRI, and explores their associations with cognitive function to help clarify the neuroimaging mechanisms underlying cognitive disorders in depression.

MATERIALS AND METHODS

Study participants

We retrospectively reviewed the medical records of Han Chinese patients with MDD treated between January 2020 and June 2024. Eligibility criteria were: (1) Diagnosis of MDD according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition or the Chinese Classification of Mental Disorders, Third Edition; (2) Age 18-45 years and right-handed; (3) 17-item Hamilton Depression Rating Scale (HAMD-17) score > 24; (4) No history of head injury, central nervous system (CNS) infection, or organic psychosis; (5) Completion of MRI without significant motion or physical illness affecting image quality; (6) No history of drug use within one month before MRI; (7) Sufficient baseline cognition to complete assessment; and (8) Complete clinical data. Exclusion criteria included: (1) History of head trauma, organic mental disorder, or CNS infection; (2) Substance-induced mental disorders; (3) Severe cardiac, hepatic, or renal dysfunction; (4) Prior bipolar disorder or schizophrenia; (5) Intellectual disability; (6) MRI contraindications; and (7) Incomplete clinical data. Based on these criteria, 86 patients formed the observation group. Additionally, 50 healthy volunteers were enrolled as controls using the same criteria except for the diagnosis.

Data collection and methods

The demographic data included age, sex, and years of education, all extracted from patients’ electronic medical records. Imaging indices comprised fractional anisotropy (FA), mean kurtosis (MK), and mean diffusivity (MD) measured across predefined brain regions.

Depression severity measurement

The HAMD-17 was used to assess depression severity, which evaluates depressed mood, anxiety, sleep disorders, loss of appetite, guilt, suicidal thoughts, and somatic symptoms on a 0-4 scale. Total scores classify depression as mild (7-16), moderate (17-24) or severe (> 24).

Cognitive function evaluation

The Montreal Cognitive Assessment (MoCA) was used to assess overall cognition, covering visuospatial/executive function, naming, attention, language, abstraction, delayed recall, calculation, and orientation. Scores range from 0 to 30 points, with > 26 indicating normal cognition; 18-26, 10-17, and < 10 correspond to mild, moderate, and severe impairment, respectively. Executive function was evaluated using the Digit Symbol Substitution Test (DSST), in which higher scores (maximum 90) indicate better performance. The Trail Making Test Part A (TMT-A) was used to assess attention, and Trail Making Test Part B (TMT-B) assessed executive function; longer completion times reflected greater impairment. All assessments were administered by trained professionals.

MRI protocol

Image acquisition: MRI was performed using a GE Sigma 3.0 T superconducting scanner with an 8-channel head coil. Conventional axial, sagittal, and coronal T1-weighted, T2-weighted, and fluid-attenuated inversion recovery sequences were obtained [slice thickness 6.0 mm, interslice gap 1.0 mm, field of view (FOV) 24 cm, averages 1-2]. DKI was acquired axially using a T2-weighted Echo-Planar Imaging sequence (repetition time 9200 milliseconds, echo time 118.3 milliseconds, FOV 24 cm, matrix size 128 × 128, 26 slices, slice thickness 5.0 mm with no gap, b values 0/1000/2000 seconds/mm², number of excitations 2, 31 diffusion directions). Total acquisition time was 9 minute 58 seconds.

Data processing: Imaging data were processed using the GE ADW 4.4 platform and Function Tool 6.0. Two senior radiologists performed quality control and parameter optimization to generate pseudocolor-coded FA, MD, and MK maps fused with axial T2 weighted imaging. Regions of interest (ROIs) (22 mm²) were placed in bilateral superior frontal gyri (SFG), middle frontal gyri (MFG), and inferior frontal gyri (IFG), temporal and occipital lobes, and the genu of the corpus callosum (GCC) and splenium of the corpus callosum (SCC). Each region underwent three repeated measurements, and mean values were calculated. ROI placement avoided cerebral sulci, vessels, and bone.

Statistical analysis

Statistical analyses were conducted using SPSS 26.0. n (%) were analyzed using χ² tests. Continuous data (mean ± SD) were compared using independent-sample t-tests. Cognitive functioning was analyzed with covariance analysis, using age, sex, and years of education as covariates. To address multiple comparisons, false discovery rate (FDR) correction was applied. Partial correlations between FA or MK and cognitive scores were calculated for regions showing significant differences, with FDR correction applied. Statistical significant was set at P < 0.05.

RESULTS

General patient information

The observation group (n = 86) included 35 men and 51 women aged 22-42 years (mean 30.77 ± 5.14) with an average of 11.19 ± 2.32 years of education (range: 7-20 years). The control group comprised 50 individuals (21 men, 29 women), aged 18-43 years (mean 31.06 ± 4.96) with 6-15 years of formal education (mean: 10.88 ± 2.09). There were no significant baseline differences between the groups (P > 0.05) (Table 1).

Table 1 Comparative analysis of general data, mean ± SD.

	Observation group (n = 86)	Control group (n = 50)	χ2/t value	P value
Sex			0.022	0.882
Male	35	21
Female	51	29
Age (years)	30.77 ± 5.14	31.06 ± 4.96	0.324	0.746
Years of education	11.19 ± 2.32	10.88 ± 2.09	0.769	0.443

HAMD scores

On the HAMD scale, the observation group scored significantly higher for depression, anxiety, sleep disorders, loss of appetite, guilt, suicidal thoughts, and physical symptoms, with all intergroup differences reaching statistical significance (P < 0.05; Table 2).

Table 2 Comparative Hamilton Depression Rating Scale assessment, mean ± SD.

	Observation group (n = 86)	Control group (n = 50)	t value	P value
Depression	3.77 ± 0.50	1.66 ± 0.66	21.04	< 0.0001
Anxiety	3.81 ± 0.39	1.82 ± 0.72	20.92	< 0.0001
Sleep disorders	3.44 ± 0.57	1.56 ± 0.58	18.43	< 0.0001
Loss of appetite	3.64 ± 0.51	1.74 ± 0.49	21.25	< 0.0001
Guilt	3.55 ± 0.55	1.20 ± 0.95	18.29	< 0.0001
Suicidal thoughts	3.47 ± 0.52	0.81 ± 0.58	27.56	< 0.0001
Physical symptoms	3.69 ± 0.27	1.08 ± 0.44	42.90	< 0.0001
Global	25.36 ± 0.65	9.90 ± 1.54	81.59	< 0.0001

Neuropsychological assessment of outcomes

The observation group had significantly lower MoCA scores than controls (t = 32.88, P < 0.0001). They also showed slower processing speeds on TMT-A (t = 5.561, P < 0.0001) and TMT-B (t = 10.31, P < 0.0001), as well as reduced DSST performance (t = 7.993, P < 0.0001; Table 3).

Table 3 Intergroup differences on neuropsychological assessments, mean ± SD.

	Observation group (n = 86)	Control group (n = 50)	t value	P value
MoCA	19.22 ± 1.88	28.30 ± 0.68	32.88	< 0.0001
TMT-A (second)	69.51 ± 17.23	53.36 ± 14.64	5.561	< 0.0001
TMT-B (second)	106.97 ± 33.27	87.90 ± 6.28	10.31	< 0.0001
DSST	57.56 ± 10.07	44.16 ± 8.19	7.993	< 0.0001

MoCA: Montreal Cognitive Assessment; TMT-A: Trail Making Test Part A; TMT-B: Trail Making Test Part B; DSST: Digit Symbol Substitution Test.

FA across brain regions

FA values in the bilateral SFG, left MFG, left IFG, occipital lobes, GCC, and SCC differed insignificantly between groups (P > 0.05). The only significant finding was reduced FA values in the bilateral temporal lobes, right MFG and right IFG of the observation group (P < 0.05; Table 4).

Table 4 Comparative analysis of fractional anisotropy across brain regions, mean ± SD.

Groups	Left superior frontal gyrus	Right superior frontal gyrus	Left middle frontal gyrus	Right middle frontal gyrus	Left inferior frontal gyrus	Right inferior frontal gyrus
Observation group (n = 86)	0.48 ± 0.10	0.46 ± 0.08	0.40 ± 0.05	0.38 ± 0.10	0.42 ± 0.08	0.44 ± 0.06
Control group (n = 50)	0.50 ± 0.05	0.48 ± 0.05	0.42 ± 0.06	0.40 ± 0.09	0.41 ± 0.06	0.46 ± 0.05
t value	1.320	0.364	2.088	1.166	0.767	1.989
P value	0.189	0.716	0.039	0.246	0.444	0.049
Groups	Left temporal lobe	Right temporal lobe	Left occipital lobe	Right occipital lobe	Genu of the corpus callosum	Splenium of the corpus callosum
Observation group (n = 86)	0.40 ± 0.06	0.38 ± 0.06	0.48 ± 0.07	0.48 ± 0.07	0.80 ± 0.09	0.80 ± 0.05
Control group (n = 50)	0.46 ± 0.04	0.50 ± 0.09	0.46 ± 0.05	0.50 ± 0.10	0.79 ± 0.09	0.79 ± 0.04
t value	6.299	9.316	1.773	1.367	0.625	1.207
P value	< 0.0001	< 0.0001	0.078	0.174	0.533	0.230

MK across brain regions

MK values differed insignificantly in the bilateral MFG, IFG, occipital lobes, GCC, or SCC (P > 0.05). In contrast, the observation group showed reduced MK values in the bilateral SFG and temporal lobes (P < 0.05; Table 5).

Table 5 Inter-group comparison of mean kurtosis values across brain regions, mean ± SD.

Groups	Left superior frontal gyrus	Right superior frontal gyrus	Left middle frontal gyrus	Right middle frontal gyrus	Left inferior frontal gyrus	Right inferior frontal gyrus
Observation group (n = 86)	0.83 ± 0.14	0.83 ± 0.10	0.92 ± 0.10	0.92 ± 0.11	0.98 ± 0.14	0.93 ± 0.09
Control group (n = 50)	0.95 ± 0.09	0.95 ± 0.09	0.91 ± 0.07	0.89 ± 0.08	0.99 ± 0.07	0.94 ± 0.07
t value	5.438	6.995	0.623	1.686	0.472	0.676
P value	< 0.0001	< 0.0001	0.534	0.094	0.638	0.501
Groups	Left temporal lobe	Right temporal lobe	Left occipital lobe	Right occipital lobe	Genu of the corpus callosum	Splenium of the corpus callosum
Observation group (n = 86)	0.92 ± 0.04	0.91 ± 0.08	1.01 ± 0.08	1.03 ± 0.06	1.00 ± 0.78	1.32 ± 0.15
Control group (n = 50)	1.03 ± 0.05	0.97 ± 0.06	1.03 ± 0.05	1.05 ± 0.06	1.13 ± 0.49	1.27 ± 0.21
t value	14.08	4.601	1.595	1.874	1.062	1.613
P value	< 0.0001	< 0.0001	0.113	0.063	0.290	0.109

MD across brain regions

MD values showed no significant differences in any region between the groups (all P > 0.05; Table 6).

Table 6 Intergroup differences in mean diffusivity across brain regions, mean ± SD.

Groups	Left superior frontal gyrus	Right superior frontal gyrus	Left middle frontal gyrus	Right middle frontal gyrus	Left inferior frontal gyrus	Right inferior frontal gyrus
Observation group (n = 86)	0.88 ± 0.05	0.87 ± 0.11	0.92 ± 0.06	0.92 ± 0.05	0.90 ± 0.07	0.92 ± 0.07
Control group (n = 50)	0.86 ± 0.10	0.84 ± 0.14	0.91 ± 0.06	0.90 ± 0.08	0.91 ± 0.06	0.90 ± 0.05
t value	1.553	1.385	0.937	1.795	0.845	1.773
P value	0.123	0.168	0.350	0.075	0.399	0.078
Groups	Left temporal lobe	Right temporal lobe	Left occipital lobe	Right occipital lobe	Genu of the corpus callosum	Splenium of the corpus callosum
Observation group (n = 86)	1.02 ± 0.09	0.95 ± 0.08	0.94 ± 0.08	0.94 ± 0.05	1.03 ± 0.12	0.94 ± 0.05
Control group (n = 50)	1.01 ± 0.10	0.94 ± 0.09	0.93 ± 0.08	0.92 ± 0.07	0.99 ± 0.19	0.93 ± 0.10
t value	0.600	0.671	0.703	1.935	1.505	0.777
P value	0.550	0.503	0.483	0.055	0.135	0.439

Correlation analysis of FA/MK and cognitive level in different regions

Correlational analyses showed no significant relationships between FA and MoCA, TMT-A, TMT-B, or DSST scores (P > 0.05). However, MK values in the bilateral SFG correlated positively with MoCA scores (P < 0.05), while left SFG MK showed a negative association with DSST performance (P < 0.05; Table 7).

Table 7 Correlations of regional fractional anisotropy and mean kurtosis with cognitive metrics.

	MoCA		TMT-A		TMT-B		DSST
	r value	P value	r value	P value	r value	P value	r value	P value
FA value
Left temporal lobe	-0.002	0.984	0.098	0.391	0.013	0.905	0.069	0.523
Right temporal lobe	0.095	0.384	-0.117	0.283	0.049	0.652	0.100	0.358
MK value
Left superior frontal gyrus	0.331	0.002	0.033	0.760	0.018	0.866	-0.265	0.014
Right superior frontal gyrus	0.422	< 0.0001	-0.117	0.284	-0.140	0.197	0.081	0.459
Left temporal lobe	-0.092	0.401	-0.089	0.413	-0.111	0.307	0.076	0.486
Right temporal lobe	0.048	0.657	0.096	0.378	0.073	0.504	-0.172	0.114

FA: Fractional anisotropy; MK: Mean kurtosis; MoCA: Montreal Cognitive Assessment; TMT-A: Trail Making Test Part A; TMT-B: Trail Making Test Part B; DSST: Digit Symbol Substitution Test.

DISCUSSION

Depression is a recurrent and progressive mental health condition often overlooked despite its substantial functional impact. Beyond persistent sadness and hopelessness, patients commonly experience marked executive dysfunction that disrupts daily activities and hinders recovery after discharge[13]. Such impairment is a major barrier to restoring social function. Prior research consistently links depression with cognitive decline, although the mechanism remains unclear. The vascular depression hypothesis supported by multiple studies proposes that isolated or cumulative small-vessel injury may disrupt prefrontal lobe circuits involved in mood and cognitive processing[14,15]. With advances in neuroimaging, structural and microstructural abnormalities have been observed in depressed patients, resembling changes seen in early vascular dementia.

This study first compared neuropsychological performance between patients with depression and healthy controls. Depressed patients showed lower MoCA scores, longer TMT-A and TMT-B completion times, and reduced digit-symbol substitution performance, indicating impairments across multiple cognitive domains and aligning with previous research. To date, only four meta-analyses[16-19] have compared cognitive deficits in clinically diagnosed MDD patients with healthy individuals, consistently identifying cognitive impairment as a core feature of MDD[20,21]. One meta-analysis reported deficits in processing speed, attention, memory, and executive functioning[17]. Evidence also shows that cognitive impairment is present during both acute episodes and remission[22], and may worsen or recur without timely treatment[23]. MRI and diffusion tensor imaging (DTI) revealed reduced bilateral temporal lobe FA values in MDD patients, suggesting disrupted white-matter microstructure. The “vascular depression” hypothesis proposes that cerebrovascular disease may precipitate or sustain depressive syndromes in older adults. One research group[24] defined this condition on the basis of MRI-detected vascular changes, particularly high-signal abnormalities. MRI studies reported significantly reduced cerebral blood flow in emotion-related regions, including the prefrontal cortex[25,26], anterior cingulate cortex[27], thalamus, and left superior temporal gyrus[28]. FA reflects the anisotropy of water-molecule diffusion; larger FA values indicate well-oriented, intact fiber tracts, whereas lower FA values reflect microstructural damage. FA ranges from 0 (isotropic diffusion) to 1 (maximal anisotropy)[29]. DTI studies on MDD have consistently shown widespread FA reductions, particularly in the frontal and temporal lobes, though few have assessed hemispheric asymmetry[30]. FA abnormalities in MDD likely reflect microstructural disruption of underlying fiber pathways. In a meta-analysis, Liao et al[31] identified reduced FA values in white matter fiber tracts connecting the prefrontal cortex to temporal, occipital, and frontal lobes, as well as to subcortical structures such as the amygdala and hippocampus, suggesting altered structural networks in MDD. We found no significant MK differences between groups in the bilateral MFG, IFG, occipital lobes, GCC, or SCC. However, patients with MDD displayed markedly reduced MK values in the bilateral SFG and temporal lobes. MK, a dimensionless metric independent of fiber-orientation effects, reflects microstructural complexity based on diffusion kurtosis. Higher MK values indicate more complex tissue architecture, whereas reduced MK suggests structural simplification or disruption[32]. MK is considered more sensitive than FA or MD in detecting microstructural changes. Consistent with frontotemporal pathology, reduced MK may reflect atrophy or demyelination of frontotemporal nerve fibers, leading to loosened white-matter organization[23]. This supports the view that white-matter disorganization stems from degenerative changes in frontotemporal circuits. Depression is understood as a disorder of brain networks, particularly those involving frontotemporal fiber bundles essential for emotional regulation[33]. The frontal lobe governs movement, judgment, planning, and emotional mood[34], whereas the temporal lobe contributes to mood regulation and other cognitive functions[35]. These findings suggest that structural alterations in the frontal and temporal lobes may appear earlier than in other regions during the course of depression. Finally, no significant MD differences were observed between MDD cases and controls across the bilateral SFG, MFG, IFG, temporal and occipital lobes, GCC, and SCC. While FA and MD assesses water molecule diffusion in the Gaussian domain, neither resolves crossing-fiber complexity.

Finally, we found no significant correlations between regional FA and MoCA, TMT-A, TMT-B, or DSST in MDD patients. In contrast, bilateral SFG MK was positively associated with MoCA scores, while left SFG MK correlated negatively with DSST performance. White matter is mainly supplied by the slender, deep-penetrating perforating arteries with short, vertical branches. As a periventricular structure, located in a vascular boundary zone, white matter is particularly vulnerable to microstructure injury when local or systemic blood flow becomes abnormal. Such changes can lead to arteriole sclerosis or occlusion and ultimately evolve into leukoaraiosis. Because the frontal lobe is mainly responsible for cognitive and executive functions, depressed patients commonly exhibit reduced cognitive control[36]. These findings support an association between cerebral white-matter changes and cognitive dysfunction in depression. However, the pattern we observed differs from the expectation that greater structural integrity should predict better cognitive performance[37]. This discrepancy may be related to the lack of differentiation between first-episode and relapsing patients, as well as variations in disease severity. Given the greater sensitivity of MK compared with FA, reductions in both MK and FA values in the temporal lobe suggest that structural alterations in the frontal and temporal lobes may precede changes elsewhere, with the temporal lobe potentially showing more pronounced abnormalities. Previous studies have not examined this hypothesis, and confirmation will require a larger sample size.

This study has two main limitations. First, the sample size was relatively small, which may weaken statistical power and limit the generalizability of the findings. Second, we did not distinguish between first-episode and recurrent depression or grade disease severity, which may explain why fewer brain regions showed significant results compared with previous studies. Further research is needed to expand the sample size and distinguish between first-episode and recurrent patients to explore depression pathogenesis. We also did not stratify patients by episode frequency or disease duration. Future work should incorporate larger cohorts and prospectively compare first-episode and recurrent cases to better clarify depression pathogenesis.

CONCLUSION

MRI remains a preferred tool for examining the relationship between white-matter structural alterations and depression pathogenesis. It can detect subtle structural abnormalities in MDD patients, and our findings indicate possible atrophy of bilateral SFG and temporal white-matter tracts, which appear to correlate with cognitive dysfunction in affected individuals.