Role of disturbed structural connectivity network in patients with white matter lesions and cognitive impairment revealed by diffusion-tensor magnetic resonance imaging

Abstract

Objective

This study aimed to compare small-world network attributes between control participants without memory complaints and patients with white matter lesions showing cognitive impairment.

Methods

Changes in diffusion tensor imaging and white matter fiber bundles in patients were analyzed. Brain structural network was constructed based on diffusion tensor imaging data, and topological properties of whole-brain small-world network were discussed. The damaged brain areas of patients with white matter lesions were studied, and the correlation between white matter lesion-related brain structural network abnormalities and cognitive impairment severity was discussed.

Results

Compared with the normal control with normal cognition group, fractional anisotropy was significantly reduced in the white matter lesions–non-dementia vascular cognitive impairment (WMLs-VCIND) and white matter lesions–vascular dementia (WMLs-VaD) groups, while mean diffusivity and radial diffusivity values were significantly increased (p < 0.05). The small-world network attributes demonstrated significant changes in λ/γ/σ values compared with the normal control with normal cognition group (p < 0.05). In the WMLs-VaD group, brain areas with reduced node-efficiency were mainly concentrated in the posterior cingulate gyrus, posterior cingulate gyrus, middle and superior lobes of right occipital region, superior lobe of the left occipital region, and right thalamus (p < 0.05). Nodal efficiencies in the WMLs-VaD group were lower than those in the WMLs-VCIND group (p < 0.05). All small-world parameters were significantly correlated with the total Mini-Mental State Examination and Montreal Cognitive Assessment scores.

Conclusions

There was extensive and subtle white matter fiber bundle damage in patients with white matter lesions. The brain structural network of patients with white matter lesions had small-world characteristics, and a reduction in small-world properties was related to reduction in cognitive function scores.

Keywords

Leukoaraiosis white matter lesions cognitive impairment magnetic resonance imaging small-world network

Introduction

Age-related white matter lesions (WMLs), also called leukoaraiosis, are very common findings in brain magnetic resonance imaging (MRI) examinations in older adults.¹ Many studies have demonstrated an association between WMLs and cognitive decline.² However, the functional status of these patients ranges from normal to severe cognitive disability, and there is no clear dose–effect relationship between the severity of WMLs and cognitive impairment. One possible reason for this is that the microstructure lesions of white matter (WM) cannot be visualized using conventional MRI. Another reason could be that WM disconnections play a more important role in cognitive decline, while conventional MRI has limited ability to visualize these.

Diffusion tensor imaging (DTI) can quantify the microstructure lesions of WM tracts.³ Fractional anisotropy (FA) and mean diffusivity (MD) are common metrics that provide a typical measure of WM alteration. The directional diffusivity measures axial diffusivity, λ1, and radial diffusivity, λ23, of WM tracts have been proposed to assess axonal injury and demyelination. Several previous DTI studies have utilized DTI parameters to investigate potential changes in patients with WMLs who show cognitive impairment.⁴ However, using region-of-interest (ROI) and tractography-based quantitative analyses cannot adequately evaluate the more widespread effects of noninvasive WMLs, especially the disruption of WM integrity.

The tract-based spatial statistics (TBSS) method applies voxel-wise statistics to analyze DTI parameters, which can minimize the tract misalignment effects. This method is now being increasingly used to study WM abnormalities in various brain diseases, such as Alzheimer’s disease (AD), schizophrenia, and multiple sclerosis (MS).

Graph theoretical analysis represents the brain as a large-scale complex network composed of brain regions and WM connectivity.⁵ Evidence has demonstrated that the brain exhibits a small-world character,⁶ which efficiently organizes information segregation and integration, optimally balancing the cost of maintaining connections and conveying information. These efficient networks are disrupted in several neurological disorders. Neurodegenerative diseases spread from region to region, adopting a network-based spatial pattern.⁷ However, the role of the WM microstructural topological organization in cognitive decline in patients with WMLs remains unclear.

Graph theoretical analysis using DTI provides quantitative information regarding the topological properties of brain networks. In addition, resting-state functional (f)MRI and graph theory studies have demonstrated disrupted small-world networks in patients with cognitive impairment. The functional plasticity of the brain is reflected in its capacity to perform anatomical reorganization. Furthermore, limited information is available regarding the alterations in the topological organization or small-world networks of the WM network (WMN) associated with cognitive impairments.^8–11 However, this type of small-world network is also correlated with cognitive impairment. Therefore, the present study aimed to compare small-world network attributes between control participants without memory complaints and WMLs patients showing cognitive impairment.

Materials and methods

Participants

This study was designed as a prospective study. Initially, 56 patients with WMLs from our hospital were enrolled in this study from January 2019 to January 2023. Among the total 56 patients with WMLs, 48 underwent the imaging procedure and completed the clinical cognitive assessment. Two participants from the WMLs group were excluded due to poor DTI data quality. Healthy controls were also recruited during the same period. In total, 36 baseline-matched healthy controls with normal cognition were included. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.¹²

The diagnosis of WMLs was made unanimously by two radiologists who independently evaluated the fluid-attenuated inversion recovery (FLAIR) magnetic resonance (MR) images visually without any knowledge of the participants’ clinical profiles. The inclusion criteria were as follows: (a) age between 50 and 85 years; (b) presence of WM hyperintensities on FLAIR MR images in accordance with the revised version of the Fazekas scale (score ≥1); and (c) presence of a contactable informant throughout the study. The exclusion criteria were as follows: (a) cardiac or renal failure, cancer, or other severe systemic diseases; (b) unrelated neurological diseases such as epilepsy, traumatic brain injury, and MS; (c) chronic cerebral infarction or other lesions; (d) leukoencephalopathy of nonvascular origin; (e) dementia of nonvascular origin; (f) psychiatric diseases or drug addiction; (g) consciousness disruption or aphasia; and (h) inability or refusal to undergo brain MRI.

Ethics statement

This study was approved by the ethics committee of our hospital. In accordance with the mandatory requirements for patients with cognitive impairment, written informed consent was obtained from their legitimate guardians. This study was conducted in accordance with the Helsinki Declaration of 1975 as revised in 2024.

Clinical cognitive assessment

All participants were instructed to complete the Chinese version of the Mini-Mental State Examination (MMSE), Beijing version of the Montreal Cognitive Assessment (MoCA), and Clinical Dementia Rating (CDR) scale under the supervision of a physician. The tests were completed in a strict order, according to the standard protocols in a quiet room. The following educational level-specific reference cutoff values for MMSE scores were used: (a) middle and high school, 27; (b) elementary school, 24; and (c) illiterate, 21. For the MoCA, the cutoff value for cognitive impairment was <26. In addition, 1 additional point was added to the raw MoCA score when the patient’s educational level was <12 years.

Based on the results of these cognitive tests, the participants were divided into three groups: (a) WML patients with nondementia vascular cognitive impairment (WMLs-VCIND, n = 30); (b) WML patients with vascular dementia (WMLs-VaD, n = 16); and (c) normal control with normal cognition (NC) (n = 36). Patients fulfilling the following criteria were classified into the WMLs-VCIND group: (a) CDR score = 0.5; (b) 24≤ MMSE score <27 with educational level ≥6 years, 20≤ MMSE <24 with educational level <6 years, or 17≤ MMSE <21 with educational level = 0 years; and (c) MoCA score <26. Patients fulfilling the following criteria were classified into the WMLs-VaD group: (a) CDR score ≥1; (b) MMSE <24 with educational level ≥6 years, MMSE <20 with educational level <6 years, or MMSE <17 with educational level = 0 years; and (c) MoCA score <22.

Brain MRI data acquisition

MRI was performed using a Siemens Magnetom Verio 3 T superconducting MRI system in the Department of Radiology. A T2-weighted (T2W)-FLAIR sequence was applied to detect WM hyperintensities. A standard T1-weighted three-dimensional (3D) magnetization-prepared rapid gradient echo sequence was applied with the following: (a) repetition time (TR), 2300 ms; (b) echo time (TE), 3.28 ms; (c) time inversion (TI), 1200 ms; (d) matrix size, 256 × 256; (e) flip angle, 9°; (f) slice thickness, 1 mm; (g) interslice gap, 0.5 mm; and (h) number of slices, 256. DTI sequences were applied with TR = 4900 ms, TE = 93 ms, voxel size = 2.5 × 2.5 × 2.5 mm; four unweighted scans, and 30 directions with b-value = 1000 s/mm². Rubber earplugs were used to reduce noise, and foam cushions were used to fix the head to minimize potential motion artifacts.

DTI preprocessing

All images obtained using DTI were preprocessed using an automated tool for diffusion MRI called pipeline for analyzing brain diffusion images (PANDA). First, the digital imaging and communications in medicine (DICOM) files were converted to Neuroimaging Informatics Technology Initiative (NIfTI) images. Next, the brain mask required for the subsequent processing steps was estimated. Third, the nonbrain space in the raw images was cutoff to reduce image size and memory cost and speed up the processing in the subsequent steps. Fourth, each diffusion-weighted image was coregistered to the b0 image by using an affine transformation to correct the eddy current-induced distortions and simple head-motion artifacts. The diffusion gradient directions were adjusted accordingly. Fifth, a voxel-wise calculation of the tensor matrix and the diffusion tensor matrices were obtained for each participant, including FA, MD, axial diffusivity (L1), and radial diffusivity (RD; L23).

Structural network construction

WM tractography

WM deterministic tractography was performed to obtain whole-brain WM tracts. Specifically, the fiber assignment continuous tracking algorithm was applied to reconstruct whole-brain tracts. To eliminate the impact of the seed point’s number and position, a single seed point was positioned at the center of each voxel of the whole brain. A tract was terminated if the turn angles of the fiber were >45° or if the fiber entered a voxel with FA <0.2. The tracking procedure was performed using the Diffusion toolkit (http://trackvis.org) embedded in PANDA.

Network node definition

In this study, we employed the automated anatomical labeling (AAL) template to parcellate the cerebral cortex into 90 cortical regions (45 for each hemisphere, Supplementary Table 1), each representing a node of the cortical network. Briefly, individual T1-weighted images were coregistered to the b0 images in the DTI space. The T1 images were then nonlinearly transformed to the T1 template of ICBM152 in the Montreal Neurological Institute (MNI) space. Inverse transformations were used to warp the AAL atlas from the MNI space to the DTI native space. Notably, AAL discrete labeling indices were preserved by using a nearest-neighbor interpolation method.

Network edge definition

To define the network edges, that is brain region pair-wise connections, a threshold value of 3 was selected for the fiber bundles, which means that two regions were considered structurally connected only if at least the two end points of three fibers were located in each of the two regions. A threshold selection can reduce the risk of false-positive connections due to noise or the limitations of deterministic tractography and can simultaneously ensure the size of the largest connected component (i.e. 90) in the networks. Bai et al.¹³ reported that this “thresholding” procedure did not significantly influence their results after they evaluated the effects of different thresholds (fiber numbers ranging from 1 to 5) on the network analysis.

After defining the network nodes and edges, binary network analyses were performed. We defined the network edges as 1 if the fiber number between the two regions was larger than the threshold (3 in our case) and as 0 if it was lower than the threshold. The result was a binary network for each participant, which was represented by a symmetric 90 × 90 matrix.

Network analyses

All topological properties were calculated, including the shortest path length (Lp), clustering coefficient (Cp), global efficiency (Eg), local efficiency (Eloc), and nodal efficiency (nodalE). The small-worldness parameters, including lambda (λ), gamma (γ), and sigma (σ), were also determined. All network analyses were performed using in-house graph theoretical network analysis (GRETNA) software (http://www.nitrc.org/projects/gretna/) and visualized using BrainNet Viewer software (http://www.nitrc.org/projects/bnv/).

Statistical analyses

Chi-square test was used to identify sex-based differences, and one-way analysis of variance (AVONA) was performed to evaluate group-based differences in age, educational level, sex, and neurological parameters. All brain network properties were compared using analysis of covariance (ANCOVA), with age and sex as covariances. Accordingly, p < 0.05 (type I error rate) was regarded to indicate a statistically significant difference. False discovery rate (FDR) correction was applied for multiple comparisons in the tract-based TBSS analyses. Subsequently, Bonferroni-corrected post-hoc analyses were performed to compare global network properties and nodalE between the 3 groups (p < 0.05/n = 0.016, where “n” represented the number of comparisons). Additionally, Pearson’s correlation analysis with FDR correction was used to evaluate the relationships between significantly altered network topological properties and behavioral performances, controlling for age and sex (p < 0.05).

Results

Demographic data and clinical variables

As shown in Table 1, there were no statistically significant differences in the age; sex; educational level; incidences of hypertension, diabetes mellitus, and dyslipidemia; or histories of smoking and alcohol consumption between the WMLs-VaD, WMLs-VCIND, and NC groups. Patients in the WMLs-VAD and WMLs-VCIND groups showed significantly lower MMSE (F = 46.42, p < 0.001) and MoCA (F = 71.96, p < 0.001) scores.

Table 1.

Demographic and clinical characteristics of patients with white matter lesions (WMLs).

Characteristics	NC(n = 36)	WMLs-VCIND(n = 30)	WMLs-VaD(n = 16)	F/t	p-value
Age (years)	66.20 ± 7.18	65.59 ± 7.59	62.83 ± 6.97	1.828	0.167
Female patients, n (%)	17 (47.22%)	14 (46.66%)	6 (37.50%)	1.682	0.431
Educational level (years)	11.3 ± 3.3	11.2 ± 3.5	11.3 ± 3.2	0.21	0.975
Hypertension, n (%)	13 (36.11%)	8 (27.27%)	10 (60.87%)	5.265	0.068
Smoking, n (%)	6 (16.67%)	5 (16.67%)	4 (25.00%)	0.514	0.773
Diabetes, n (%)	4 (11.11%)	2 (6.25%)	4 (25.00%)	1.364	0.505
Hyperlipemia, n (%)	5 (13.89%)	3 (10.00%)	6 (37.50%)	1.094	0.579
BMI	23.87 ± 1.18	24.27 ± 1.12	24.13 ± 1.01	0.772	0.466
MMSE	29.528 ± 0.69	27.000 ± 2.28	23.750 ± 4.32	89.89	<0.001
MoCA	27.694 ± 1.4	22.500 ± 2.40	17.87 ± 4.17	140.407	<0.001

WMLs-VCIND: white matter lesions with nondementia vascular cognitive impairment; WMLs-VaD: white matter lesions with vascular dementia; NC: normal controls with normal cognition; MMSE: Mini-Mental State Examination; MoCA: Montreal Cognitive Assessment.

Tract-based TBSS analyses

Statistical analyses revealed significantly increased MD and λ23 values in several WM tracts in patients with WMLs compared with NCs, including the bilateral corticospinal tract, frontal occipital anterior tract, and anterior thalamic radiation (p < 0.05, FDR-corrected among the 3 groups, Figure 1(a) and (b)). There were no significant tract-specific λ1 values between the groups.

Figure 1.

Group differences in (a) regional MD, (b) regional λ23, and (c) regional FA. Green represents the average white matter skeleton of all participants. The red color represents the brain region with increased FA value. Compared with the NC group, patients with WMLs exhibit significantly increased MD and fractional anisotropy (λ23) in multiple white matter tracts, including bilateral corticospinal tracts, frontal occipital frontoparietal tracts, and anterior thalamic radiation tracts (p < 0.05, FWE corrected for multiple comparisons, Figure 1(a) and (b)). Conversely, WML patients show markedly reduced FA in several white matter tracts, such as bilateral corticospinal tracts, frontal occipital frontoparietal tracts, anterior thalamic radiation tracts, and superior tracts (p < 0.05, FWE corrected for multiple comparisons, Figure 1(c)).

FA was significantly lower in several WM tracts in patients with WMLs compared with NCs, including the bilateral corticospinal tract, frontal occipital anterior tract, anterior thalamic radiation, and the superior fasciculus (p < 0.05, FDR-corrected among the 3 groups, Figure 1(c)).

Group differences in global network properties

The whole WM brain network of both WML patients and NCs exhibited small-world networks. With the increase in sparsity, λ decreases gradually, while γ increases gradually. Compared with the random network, all the networks conform to the small-world characteristics of λ ≈1, γ >1, and σ >> 1.

Compared with the NCs, WMLs-VCIND and WMLs-VaD patients exhibited significant decreases in degree (F = 6.02, p = 0.004), Eloc (F = 4.71, p = 0.012), Cp (F = 3.31, p = 0.042), and σ (F = 7.38, p = 0.001) and significant increases in Lp (F = 3.25, p = 0.044). A Bonferroni-corrected post-hoc test showed that compared with the NC and WMLs-VCIND groups, the WMLs-VaD group exhibited significant decreases in degree, Cp, and σ. No significant differences were found between the WMLs-VCIND and NC groups (p < 0.05) (Table 2 and Figure 2(a)).

Table 2.

Group differences in global network properties.

Sparsity properties	mean ± SD			ANCOVA		Bonferroni-corrected post-hoc p-value
	NC(n = 36)	WMLs-VCIND(n = 30)	WMLs-VaD(n = 16)	F	p	WMLs-VCINDvs. WMLs-VaD		WMLs-VCINDvs. NC		WMLs-VaDvs. NC
	NC(n = 36)	WMLs-VCIND(n = 30)	WMLs-VaD(n = 16)	F	p	t	p	t	p	t	p
Degree	7.325 ± 1.053	7.300 ± 1.290	6.047 ± 1.798	6.02	0.004	−2.89	0.008	−0.020	0.997	−2.97	0.005
Eg	0.401 ± 0.052	0.407 ± 0.036	0.356 ± 0.072	5.55	0.006	−3.19	0.006	−0.19	0.886	−1.42	0.014
Eloc	0.555 ± 0.066	0.550 ± 0.083	0.483 ± 0.106	4.71	0.012	−2.26	0.027	−0.054	0.958	−1.48	0.012
Lp	2.580 ± 0.723	2.479 ± 0.246	2.929 ± 0.656	3.25	0.044	−2.127	0.037	−0.31	0.757	1.51	0.118
Cp	0.387 ± 0.039	0.381 ± 0.050	0.352 ± 0.052	3.31	0.042	−1.365	0.106	0.179	0.872	−2.120	0.036
Lambda	1.102 ± 0.021	1.097 ± 0.020	1.140 ± 0.062	10.15	<0.001	−3.717	<0.001	0.23	0.806	−3.34	0.001
Gamma	3.918 ± 0.680	3.849 ± 0.509	4.913 ± 1.594	8.72	<0.001	−3.46	0.001	−0.064	0.948	−3.22	0.001
Sigma	3.548 ± 0.569	3.507 ± 0.430	4.261 ± 1.164	7.38	0.001	−3.199	0.002	−0.018	0.968	−3.07	0.002

ANCOVA: analysis of covariance; NC: normal controls with normal cognition; WMLs-VCIND: white matter lesions with non-dementia vascular cognitive impairment; WMLs-VaD: white matter lesions with vascular dementia; Eg: global efficiency; Eloc: local efficiency Lp: path length; Cp: clustering coefficient.

Figure 2.

Group differences in (a) global network properties and (b) nodal network properties. The following six brain regions exhibit significant reductions in nodal efficiency in WML patients in compared with that in controls: posterior cingulate gyrus, middle lobe of the right occipital region, superior lobe of the left occipital region, superior lobe of the right occipital region, and right thalamus. WML: white matter lesions.

Alterations in nodal network properties

Following the discovery of a disrupted global network organization, the following six brain regions exhibited significant reductions in nodalE (p < 0.05, FDR-corrected) in WMLs-VaD patients compared with that in NCs: (a) posterior cingulate gyrus; (b) middle lobe of the right occipital region; (c) superior lobe of the left occipital region; (d) superior lobe of the right occipital region; and (e) right thalamus. Bonferroni-corrected post-hoc tests identified nodal network properties with significant differences between each pair of the 3 groups (p < 0.05/3 = 0.016). No significant difference was found between the WMLs-VCIND and NC groups (Table 3 and Figure 2(b)).

Table 3.

Group differences in nodal efficiency.

Regions	NodalE (mean ± SD)			ANCOVA		Bonferroni-corrected post-hoc p-value
	NC(n = 36)	WMLs-VCIND(n = 30)	WMLs-VaD(n = 16)	F	p	WMLs-VCINDvs. WMLs-VaD		WMLs-VCINDvs. NC		WMLs-VaDvs. NC
	NC(n = 36)	WMLs-VCIND(n = 30)	WMLs-VaD(n = 16)	F	p	t	p	t	p	t	p
Cingulum_Post_R	0.384 ± 0.038	0.381 ± 0.050	0.287 ± 0.156	9.541	<0.001	−2.024	0.004	−0.29	0.795	−3.376	0.001
Cingulum_Post_L	0.389 ± 0.045	0.387 ± 0.056	0.290 ± 0.155	9.282	<0.001	−2.020	0.004	0.139	0.854	−3.385	0.001
Occipital_Mid_R	0.413 ± 0.041	0.404 ± 0.049	0.347 ± 0.078	9.108	<0.001	−2.023	0.004	−0.721	0.446	−3.768	<0.001
Occipital_Sup_L	0.462 ± 0.051	0.464 ± 0.039	0.400 ± 0.078	8.704	<0.001	−3.376	0.001	0.159	0.893	−3.378	0.001
Occipital_Sup_R	0.474 ± 0.051	0.460 ± 0.043	0.405 ± 0.083	8.496	<0.001	−2.034	0.005	−1.242	0.254	−3.416	0.001
Thalamus_R	0.439 ± 0.053	0.434 ± 0.059	0.371 ± 0.068	8.262	0.001	−3.199	0.002	0.281	0.760	−3.996	<0.001

ANCOVA: analysis of covariance; NC: normal controls with normal cognition; WMLs-VCIND: white matter lesions with nondementia vascular cognitive impairment; WMLs-VaD: white matter lesions with vascular dementia; NodalE: nodal efficiency.

Relationship between small-world network properties and cognitive scores

As shown in Table 4, when all participants were included in one large group, all small-world parameters were significantly correlated with the total MMSE and MoCA scores (Figure 3(a) and (b)).

Table 4.

Relationship between small-world network properties and cognitive scores.

Variables		MMSE	MoCA
Degree	r	0.350	0.403
Degree	p	0.001	<0.001
Eg	r	0.336	0.361
Eg	p	0.002	0.001
Eloc	r	0.316	−0.280
Eloc	p	0.004	0.011
Cp	r	0.275	−0.415
Cp	p	0.012	<0.001
Gamma	r	−0.415	−0.427
Gamma	p	<0.001	<0.001
Sigma	r	−0.382	−0.397
Sigma	p	<0.001	<0.001

Eg: global efficiency; Eloc: local efficiency Lp: path length; Cp: clustering coefficient; MMSE: Mini-Mental State Examination; MoCA: Montreal Cognitive Assessment.

Figure 3.

Relationship between small-world network properties and MMSE cognitive scores (a) and MoCA cognitive scores (b). All small-world parameters were significantly correlated with the total scores of MMSE and the total scores of MoCA. MMSE: Mini-Mental State Examination; MoCA: Montreal Cognitive Assessment.

Relationship between nodalE and cognitive scores

NodalE values in brain regions were significantly correlated with the total MMSE and MoCA scores (Table 5). Posterior cingulate gyrus (left cingulate gyrus), posterior cingulate gyrus (right cingulate gyrus), middle lobe of the right occipital region, and superior lobe of the right occipital region were positively correlated with the total MMSE and MoCA scores. The nodalE of the left superior occipital lobe and right thalamus was negatively correlated with the total MMSE and MoCA scores (Figure 4(a) and (b)).

Table 5.

Relationship between nodal efficiency and cognitive scores.

Regions		MMSE	MoCA
Cingulum_Post_R	r	0.432	0.476
Cingulum_Post_R	p	<0.001	<0.001
Cingulum_Post_L	r	0.323	0.397
Cingulum_Post_L	p	0.003	<0.001
Occipital_Mid_R	r	0.392	0.434
Occipital_Mid_R	p	<0.001	<0.001
Occipital_Sup_L	r	0.476	0.411
Occipital_Sup_L	p	<0.001	<0.001
Occipital_Sup_R	r	0.396	0.433
Occipital_Sup_R	p	<0.001	<0.001
Thalamus_R	r	0.347	0.326
Thalamus_R	p	0.001	0.003

MMSE: Mini-Mental State Examination; MoCA: Montreal Cognitive Assessment.

Figure 4.

Relationship between nodal efficiency and MMSE cognitive scores (a) and MoCA cognitive scores (b). All nodal efficiencies were significantly correlated with the total scores of MMSE and the total scores of MoCA. MMSE: Mini-Mental State Examination; MoCA: Montreal Cognitive Assessment.

Discussion

In the present study, we investigated global WM changes in patients with WMLs using TBSS. Then, using graph theory methods, we explored the characteristics of the small-world attributes of brain structural networks. We found that patients with WMLs showed significantly increased λ23 and MD values and decreased FA in several WM tracts. Meanwhile, the altered small-world properties were reduced in WMLs-VaD patients. Moreover, these network abnormalities were related to cognitive scores. These results can improve our understanding of the neuropathological effects of disrupted WM structural connectivity in patients with cognitive impairments.

In our study, the WM tracts of patients with WMLs exhibited significant changes on TBSS analysis. In comparison with controls, patients with WMLs exhibited significantly increased MD and λ23 values in several WM tracts. A previous study¹⁴ used the fiber bundle analysis method based on the atlas fiber bundle analysis method or the area of interest (ROI) to confirm the dispersion anomalies of specific WM fiber bundles, including the whole corpus callosum and bilateral cingulate tracts, consistent with our TBSS findings. Our study confirmed the existence of more extensive and subtle WM structure abnormalities in patients with WMLs, suggesting the presence of more complex pathological mechanisms in addition to demyelination. The decrease in the FA value may be related to the decrease in the water diffusion barrier function caused by axonal loss in the WML area. It is suggested that decreased FA values reflect damage to the microstructural integrity of the WM fiber bundle.¹⁵ However, changes in the MD values showed an opposite trend to those observed for FA values. Lower MD values indicated better integrity of the WM fiber. In this study, the MD value of the WML group significantly increased, that is, the WM integrity of WML group was damaged, and the WM microstructure in the brain was damaged. Compared with the NC group, WMLs-VAD and WMLs-VCIND groups showed significantly lower FA values and higher MD values, demonstrating WM microstructure damage in WML patients. However, there were no significant differences between the WMLs-VAD and WMLs-VCIND groups, suggesting comparable levels of damage to the cerebral WM microstructure.

The results of this study showed that the brain areas with abnormal WM fiber bundles in the WMLs-VaD group were mainly concentrated in the corpus callosum radiating from the occipital part, left upper and lower longitudinal tracts, frontal and occipital tracts, and left corticospinal tract. The decrease in executive function was related to damage to the bilateral upper and lower longitudinal tracts, bilateral corticospinal tracts, and corpus callosum.¹⁶ A previous study that investigated DTI abnormalities in the brain areas of WML patients has drawn different conclusions. However, consistent with most studies,¹⁷ our study found abnormalities in the corpus callosum, which, as the largest joint fiber, can coordinate the information connection of the whole brain and is an important pathway that contributes toward maintaining normal cognitive function. A previous study¹⁸ has shown that the corpus callosum may be involved in the early stage of cognitive impairment, and changes in its microstructure are an important cause of cognitive impairment.

Consistent with previous studies,¹⁹ our study showed that the brain structural network of WML patients had small-world network characteristics. The human brain is considered a complex and efficient neural network with the ability to process information with the largest clustering coefficient but the smallest shortest Lp.²⁰ This confirms that the small-world attributes of the brain structural network are altered in patients with WML-related cognitive dysfunction, indicating loss of optimized network attributes.

In this study, compared with the NC group, the Lp of the WML group was longer. Lp is considered a measure of the information processing ability of the whole-brain network for information integration between different cortical regions. The interaction of information between interconnected brain regions is considered fundamental to human cognitive processes.²¹ The longer Lp in WML patients may indicate that the information processing between the distal regions of the brain is disturbed, providing additional evidence for the mechanism of cognitive impairment. In comparison with the NC group, the Eg of WML patients also decreased significantly. Eg is a measure of the overall capability or efficiency of parallel information transmission and integrated processing in a network. Our findings suggest that the decrease in overall efficiency is the result of impaired information transmission between brain structural networks due to WM damage.

NodalE reflects the role of nodes in information processing.²² In this study, The NodalE of WMLs-VCIND and WMLs-VaD patients changed in the cortical and subcortical areas. Moreover, in WML patients, the nodalE of six key brain regions was also significantly reduced, which was significantly correlated with the cognitive function score. These six brain regions were mainly concentrated in the cingulate gyrus, occipital lobe, and thalamus; the cingulate gyrus is located in the medial lobe, which is one of the core regions of the default network²³ that participates in a variety of brain functions, including memory and executive function. The thalamus plays a key role in memory circuits. Patients with AD and mild cognitive impairment (MCI) may have decreased thalamic volume or abnormal function.²⁴ Thus, the above brain abnormalities may have some functional basis in the pathophysiological mechanism underlying cognitive decline in WML patients. More behavioral and neuroimaging studies should be conducted to verify this conclusion.

Notably, we found that the clinical cognitive scores were correlated with the topological properties (FC, nodalE, Eg, Lp, γ, σ). Cognitive impairment was associated with weaker connectivity, lower global and Eloc, longer absolute Lp, and impaired small-worldness.

Impaired small-worldness in an aging population is considered a consequence of reduction in the sensitivity of the neurons receiving the stimuli.²⁵ In addition, WMLs can negatively affect the optimal balance between information transmission and functional integration.²⁶ Impaired small-worldness in WMLs suggest an interruption of local specialization due to WM damage, which is in concordance with the theory that small-world topology appears when networks are evolved for an optimal balance between local specialization and global integration.²⁶ The small-world network model reflects an optimal balance between local specialization and global integration. In addition, our results revealed significant correlations between small-world properties and MoCA, but not MMSE scores, which is in line with previous findings.²⁷ These findings indicate the potential constructive reorganization of brain networks secondary to WMLs,²⁸ which provides novel insights into the role of small-world properties in the cognitive dysfunction in WMLs.

Despite these promising findings, there were certain limitations to our study. In addition to evaluations of structural brain networks, future research is warranted to obtain an integrated view of the structural and functional brain network changes and their associations with cognition in patients with WMLs.

Conclusion

This study used TBSS analysis based on DTI data and found that patients with WMLs had extensive and subtle WM fiber bundle damage, and the major brain areas covered by the damage indicate the anatomical basis for cognitive damage. At the same time, the graph theory analysis method based on DTI data showed that the brain structural network of WMLs patients had small-world characteristics, and changes in these characteristics were related to the cognitive function scores of patients with WMLs. These findings help understand the influence of WML on cognition from the aspect of the whole-brain topological attributes of brain network.

Supplemental Material

sj-pdf-1-imr-10.1177_03000605261431456 - Supplemental material for Role of disturbed structural connectivity network in patients with white matter lesions and cognitive impairment revealed by diffusion-tensor magnetic resonance imaging

Supplemental material, sj-pdf-1-imr-10.1177_03000605261431456 for Role of disturbed structural connectivity network in patients with white matter lesions and cognitive impairment revealed by diffusion-tensor magnetic resonance imaging by Xiaojin Ning, Min Wu, Jianjun Tan, Meina Feng, Qin Zhou, Zhimin Fan, Ling Wan, Li Zhang and Jinfang Wang in Journal of International Medical Research

Footnotes

Acknowledgments

None.

Author contributions

Xiaojin Ning and Min Wu were responsible for data collection, figures and tables preparation, data analysis, and manuscript writing. Jianjun Tan, Meina Feng, Qin Zhou, Zhimin Fan, Ling Wan, and Li Zhang were responsible for data collection, literature review, and manuscript revision. Jinfang Wang and Yumei Zhang were responsible for the study design, data explanation, and critical revisions.

All authors have read and approved the manuscript.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declaration of conflicting interests

There are no conflicts of interests regarding the publication of this study.

Ethics statement

This study was approved by the ethics committee of Wuhan Brain Hospital, General Hospital of the Yangze River Shipping, China.

Funding

This study was supported by The Funding for Scientific Research Projects from Wuhan Municipal Health Commission (Grant No. WX23Q25) and The Funding for Science and Technology Projects from the Yangtze River Shipping Administration Bureau (Grant No. 2025-CHKJ-015).

Patient consent statement

In accordance with the mandatory requirements for patients with cognitive impairment, written informed consent was obtained from their legitimate guardians.

Supplemental material

Supplemental material for this article is available online.

ORCID iD

Jinfang Wang

References

Garo-Pascual

Zhang

Valentí-Soler

, et al. Superagers resist typical age-related white matter structural changes. J Neurosci 2024; 44: e2059232024.

Quan

Lin

, et al. Adiponectin levels are associated with white matter lesions (WMLs) and cognitive impairment. Biomed Res Int 2022; 2022: 9943250.

Leonard

Renard

Harsan

, et al. Diffusion tensor imaging reveals diffuse white matter injuries in locked-in syndrome patients. PLoS One 2019; 14: e0213528.

Wang

Yuan

Guo

, et al. Correlation between prefrontal-striatal pathway impairment and cognitive impairment in patients with leukoaraiosis. Medicine (Baltimore) 2017; 96: e6703.

Long

Bhinge

Calhoun

, et al. Graph-theoretical analysis identifies transient spatial states of resting-state dynamic functional network connectivity and reveals dysconnectivity in schizophrenia. J Neurosci Methods 2021; 350: 109039.

Gong

Concha

, et al. Mapping anatomical connectivity patterns of human cerebral cortex using in vivo diffusion tensor imaging tractography. Cereb Cortex 2009; 19: 524–536.

McMackin

Muthuraman

Groppa

, et al. Measuring network disruption in neurodegenerative diseases: new approaches using signal analysis. J Neurol Neurosurg Psychiatry 2019; 90: 1011–1020.

Brown

Jiang

Smith

, et al. Age and Alzheimer’s pathology disrupt default mode network functioning via alterations in white matter microstructure but not hyperintensities. Cortex 2018; 104: 58–74.

Vergoossen

LWM

Jansen

JFA

Van Sloten

, et al. Interplay of white matter hyperintensities, cerebral networks, and cognitive function in an adult population: diffusion-tensor imaging in the Maastricht study. Radiology 2021; 298: 384–392.

10.

Taylor

Payne

Krishnan

, et al. Evidence of white matter tract disruption in MRI hyperintensities. Biol Psychiatry 2001; 50: 179–183.

11.

Pelletier

Periot

Dilharreguy

, et al. Age-related modifications of diffusion tensor imaging parameters and white matter hyperintensities as inter-dependent processes. Front Aging Neurosci 2016; 7: 255.

12.

Von Elm

Altman

Egger

; STROBE Initiativeet al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med 2007; 147: 573–577.

13.

Bai

Shu

Yuan

, et al. Topologically convergent and divergent structural connectivity patterns between patients with remitted geriatric depression and amnestic mild cognitive impairment. J Neurosci 2012; 32: 4307–4318.

14.

Chanraud

Zahr

Sullivan

, et al. MR diffusion tensor imaging: a window into white matter integrity of the working brain. Neuropsychol Rev 2010; 20: 209–225.

15.

Alfaro

Gavrieli

Saade-Lemus

, et al. White matter microstructure and cognitive decline in metabolic syndrome: a review of diffusion tensor imaging. Metabolism 2018; 78: 52–68.

16.

Hsieh

Chang

Chuang

, et al. End-stage renal disease: in vivo diffusion-tensor imaging of silent white matter damage. Radiology 2009; 252: 518–525.

17.

Sali

Verganelakis

Gotsis

, et al. Diffusion tensor imaging (DTI) in the detection of white matter lesions in patients with mild cognitive impairment (MCI). Acta Neurol Belg 2013; 113: 441–451.

18.

Stricker

Salat

Kuhn

, et al. Mild cognitive impairment is associated with white matter integrity changes in late-myelinating regions within the corpus callosum. Am J Alzheimers Dis Other Demen 2016; 31: 68–75.

19.

Watts

Strogatz

SH.

Collective dynamics of ‘small-world’ networks. Nature 1998; 393: 440–442.

20.

Wang

Individual T1-weighted/T2-weighted ratio brain networks: small-worldness, hubs and modular organization. Int J Mod Phys C 2018; 29: 1840007.

21.

Wendiggensen

Beste

How intermittent brain states modulate neurophysiological processes in cognitive flexibility. J Cogn Neurosci 2023; 35: 749–764.

22.

Balkenhol

Händel

Biswas

, et al. Beyond-local neural information processing in neuronal networks. Comput Struct Biotechnol J 2024; 23: 4288–4305.

23.

Ofer

Jacobs

Jaiser

, et al. Cognitive and behavioral comorbidities in Rolandic epilepsy and their relation with default mode network’s functional connectivity and organization. Epilepsy Behav 2018; 78: 179–186.

24.

Chen

Song

; Alzheimer’s Disease Neuroimaging Initiativeet al. Altered structural and functional homotopic connectivity associated with the progression from mild cognitive impairment to Alzheimer’s disease. Psychiatry Res 2023; 319: 115000.

25.

Achard

Bullmore

Efficiency and cost of economical brain functional networks. PLoS Comput Biol 2007; 3: e17.

26.

Vallet

Hudon

Simard

, et al. The disconnection syndrome in the Alzheimer’s disease: the cross-modal priming example. Cortex 2013; 49: 2402–2415.

27.

Chen

Rosa-Neto

, et al. Revealing modular architecture of human brain structural networks by using cortical thickness from MRI. Cereb Cortex 2008; 18: 2374–2381.

28.

Latora

Marchiori

Efficient behavior of small-world networks. Phys Rev Lett 2001; 87: 198701

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.14 MB