Vascular Effects on Depressive Symptoms in Cognitive Impairment

Abstract

Late life depression is related to pathologic burdens, such as cerebral small vascular disease (CSVD) and amyloid, which are associated with brain network changes and cortical thinning. To examine the associations of various CSVD imaging markers, amyloid, and network changes with depression in cognitively impaired patients, we prospectively recruited 228 cognitively impaired patients having various degrees of amyloid and CSVD who underwent diffuse tensor image and PiB PET. Greater CSVD burden was associated with greater Geriatric Depression Scale (GDS) (white matter hyperintensities, WMH: p = 0.025, lacunes: p < 0.001) but not with amyloid (p = 0.095), and cortical thinning (p = 0.630) was not associated with greater GDS. The changes in white matter networks were related to GDS with decreasing integration (global efficiency: p < 0.001) and increasing segregation (clustering coefficient: p = 0.009). The network changes mediated the relationships of WMH and lacunes with GDS. Our findings provide insight to better understand how CSVD burdens contribute to depression in cognitively impaired patients having varying degrees of amyloid and vascular burdens.

Keywords

Depression network small vessel disease

INTRODUCTION

Late-life depression (LLD) has been shown to be associated with higher rates of cerebrovascular disease (CVD) risk factors and CVD events [1 –4]. In this regard, the concept of vascular depression has emerged [3, 5]. Specifically, researchers have suggested that cerebrovascular disease may predispose, precipitate, or perpetuate depressive syndrome [6].

Recent studies have also suggested that amyloid is associated with depression [9, 10]. In vivo amyloid PET and pathologic studies have reported that LLD patients have more amyloid deposition than do those without depression. According to the amyloid model of LLD, amyloid deposition affects frontolimbic and frontostriatal dysfunction, which leads to LLD [10]. In patients with amyloid and vascular burdens, network changes and cortical atrophy are emerging as downstream imaging markers. That is, pathologic imaging markers, such as amyloid and vascular burdens, independently affect network changes and cortical atrophy, eventually resulting in cognitive impairments. Recent studies have also shown that depression was associated with network changes and cortical atrophy. Furthermore, previous studies suggested that vascular depression might be explained by the disconnection hypothesis. That is, vascular damage, including white matter changes, affects the brain network, which in turn leads to clinical depressive symptoms [7, 8]. Thus, to reveal the pathomechanism of LLD, it is worthwhile to address the complex relationships among pathologic imaging markers, downstream imaging markers, and depression.

In this study, we investigate these complex relationships in cognitively impaired patients having various degrees of amyloid and CSVD. We hypothesized that CSVD burden, including WMH, lacunes, and microbleeds, contributes more to depressive symptoms than amyloid burden, based on a meta-analysis study suggesting that LLD is more strongly associated with vascular dementia than Alzheimer’s disease (AD) [11].

We further hypothesized that network changes and cortical atrophy are related to depressive symptoms, which mediate the effects of pathologic imaging markers on depressive symptoms.

METHODS

Participants

We recruited 251 cognitively impaired patients (45 amnestic mild cognitive impairment (aMCI), 67 subcortical vascular MCI (svMCI), 69 AD, and 70 subcortical vascular dementia (SVaD)) who underwent both brain magnetic resonance imaging (MRI) and Pittsburgh compound B (PiB) positron emission tomography (PET) between July 2007 and July 2011 at Samsung Medical Center. The MCI patients met Petersen’s criteria [12] with the following modifications, as previously described [13]: 1) subjective cognitive complaint by the patient or his/her caregiver, 2) objective memory decline below the 16th percentile on neuropsychological tests [14], 3) normal general cognitive function above the 16th percentile on the Korean version of the Mini–Mental State Examination (MMSE), 4) normal activities of daily living (ADL) as judged by both an interview with a clinician and the standardized ADL scale previously described [15], and 5) not demented. Of those who met the MCI criteria, aMCI was further diagnosed when patients showed objective memory decline on neuropsychological tests (below the 16th percentile of age- and education- matched norms on visual memory or verbal memory tests as described below). AD patients met the diagnostic criteria for probable AD dementia according to the National Institute of Neurological and Communicative Disorders and Stroke and the AD and Related Disorders Association (NINCDS-ADRDA) [16]. Patients with svMCI were diagnosed using the Petersen criteria [12] with modifications [13]. SVaD patients met the diagnostic criteria for vascular dementia, as determined by the Diagnostic and Statistical Manual of Mental Disorders–Fourth Edition (DSM-IV) and also met the imaging criteria for SVaD, as proposed by Erkinjuntti et al. [17] All svMCI and SVaD patients had severe WMH defined as a cap or band (periventrivular WMH or PWMH)≥10 mm and deep white-matter lesions (deep WMH or DWMH)≥25 mm, as modified from Fazekas’ MRI ischemia criteria [18]. All aMCI and AD had minimal (PWMH<5 mm and DWMH<10 mm) or moderate (between minimal and severe grades) WMH.

We excluded subjects with territorial infarction, WMH with multiple sclerosis, leukodystrophy, radiational injury, or vasculitis. Clinical interviews, neurological examinations, and blood tests were conducted with all patients. The blood test was performed to check the complete blood count, blood chemistry, syphilis serology, thyroid function, vitamin B₁₂, folate level, and apolipoprotein E (APOE) genotyping.

Of the 251 subjects, we excluded 23 patients: 4 for whom Geriatric Depression Scale (GDS) data were not available, 6 for whom WMH volume measurement failed due to segmentation errors, and 13 for whom the quality of the diffusion image (low signal-to-noise ratio) was not sufficient to reconstruct reliable fiber tracts. Thus, network analysis was performed with 228 subjects (Fig. 1).

Fig. 1.

Flowchart for participant selection. aMCI, amnestic mild cognitive impairment; AD, Alzheimer’s disease; svMCI, subcortical vascular mild cognitive impairment; SVaD, subcortical vascular dementia; WMH, white matter hyperintensities.

Standard protocol approvals, registrations, and patient consents

This study was approved by the Institutional Review Board of Samsung Medical Center. We obtained written consent from each patient.

PiB-PET acquisition and data analysis

Patients underwent a [¹¹C] PiB PET scan at Samsung Medical Center or Asan Medical Center. All patients underwent the same type of PET scan with a Discovery STe PET/CT scanner (GE Medical Systems, Milwaukee, WI, USA). Cerebellar grey matter was used as the reference region for measuring the PiB standardized uptake value ratio (SUVR). PiB SUVR was considered as a continuous variable representing the amyloid burden. The detailed radio chemistry profiles, scanning protocol, and PiB-PET data analysis were described in a previous study [19].

MR imaging techniques

We acquired standardized T2, three-dimensional T1 turbo field echo, three-dimensional fluid-attenuated inversion recovery (FLAIR), and diffusion tensor imaging (DTI) for all participants at Samsung Medical Center using the same 3.0T MRI scanner (Philips 3.0T Achieva). Detailed imaging parameters are described previously [20]. In whole-brain DT-MRI examination, sets of axial diffusion-weighted single-shot echo-planar images were collected with the following parameters: 128×128 acquisition matrix, 1.72×1.72×2 mm³ voxel size, 70 axial slices, 22×22 cm² field of view, echo time (TE) 60 ms, repetition time (TR) 7696 ms, flip angle 90°, slice gap 0 mm, and b-factor of 600 smm^–². Diffusion-weighted images were acquired from 45 different directions using the baseline image without weighting [0,0,0]. All axial sections were acquired parallel to the anterior commissure-posterior commissure line.

Measurement of WMH volume and lacunes

We measured the volume of white matter hyperintensity (WMH) (in ml) on FLAIR images using an automated method [25]. A lacune was defined as a lesion≥3 mm and≤15 mm in diameter with a low signal on T1-weighted images, high signal on T2-weighted images, and perilesional halo on FLAIR images. This meets the definition of a lacune of presumed vascular origin, as recently proposed by Wardlaw et al. [26] Two neurologists manually counted the number of lacunes, with a kappa value of 0.78. We excluded lacune volumes from total WMH volumes.

Network node definition

The T1-weighted images were registered to the ICBM152 T1 template nonlinearly in Montreal Neurological Institute (MNI) space [27]. The inverse transformation with a nearest-neighbor interpolation method was applied to the automated anatomical labeling (AAL) atlas [28], which is transformed from MNI space to T1 native space. Using the AAL atlas, we obtained 78 areas of cerebral cortex (39 regions for each hemisphere) and defined them as the nodes of the brain network.

WM tractography

Using the diffusion toolbox of the FSL package (http://www.fmrib.ox.ac.uk/fsl/fdt), distortions caused by simple head motions and eddy currents were corrected. Diffusion tensor models were estimated, and we calculated fractional anisotropy (FA) and mean diffusivity (MD) voxel-by-voxel. The WM fiber tracts in native diffusion space for each subject were reconstructed by a continuous tracking algorithm [29]. Tracking was terminated when the angle between two adjacent vectors was greater than 45° or when both ends of the fibers reached non WM, outside of the WM mask for tissue segmentation [30, 31]. Fibers shorter than 20 mm and longer than 200 mm were filtered out. WMH can be associated with alterations in WM organization; thus, we did not consider fibers that could terminate early due to WMH in network edge definition.

Network edge definition

Two nodes were connected by an edge when at least the end points of three fiber tracts were in these two nodes. A threshold number of fiber tracts was applied to reduce the risk of false-positive connections due to noise or limitations of deterministic tractography [31, 32] The mean FA value along the fibers connecting a pair of regions was used to weight the edge. Finally, we constructed weighted WM networks represented by symmetric 78×78 matrices for each individual.

Network analysis

Graph theory-based analyses were applied on the weighted WM networks using the Brain Connectivity Toolbox (http://www.brain-connectivity-toolbox.net) [33]. We used diffusion tensor imaging (DTI) data and applied graph theoretical analysis to evaluate the relationship between vascular changes and depressive symptoms from a brain network perspective. Graph theoretical analysis using DTI data can be used to examine large-scale WM connectivity in the brain network [21]. Previous studies from our group suggested that CSVD is closely correlated with WM network changes in SVCI patients [20 , 22–24].

For measures of functional integration, we calculated the shortest path length and global efficiency. The average shortest path length between two nodes in a network is known as the characteristic path length of the network [34]. The global efficiency was computed as the average inverse shortest path length [35]. For measures of functional segregation, we calculated the weighted clustering coefficient and transitivity [36]. The clustering coefficient is the fraction of triangles around each node, and the transitivity is a classical variant of the clustering coefficient [33]. A module is a group of nodes strongly connected to other nodes within the module but having weak connections to nodes outside the module. Modularity was estimated by maximizing the ratio of within-modular to between-modular edges with optimization algorithms, and the numbers of modules was measured [37].

Cortical thickness data analysis

T1-weighted images were processed using the standard MNI anatomic pipeline. The measurement process was described in a previous study [20].

Neuropsychological tests

Participants underwent neuropsychological tests using a standardized neuropsychological battery [38]. MMSE and clinical dementia rating-sum of boxes (CDR-SB) scores were used to represented cognitive states. The GDS was used to measure severity of depressive mood.

Statistical analysis

To evaluate the associations of WMH volume, number of lacunes, and white matter network parameters with GDS, we performed multiple linear regression analysis entering age, sex, and PiB SUVR as covariates. Because WMH volumes were not normally distributed, we used log_e transformed (WMH volume +1) as the exposure. Three outliers were removed.

To evaluate the association of cortical thickness with GDS, we performed multiple linear regression analysis. Covariates included age, sex, WMH volume, and intracranial volume (ICV), with ICV defined as the sum of the gray matter, white matter, and cerebrospinal fluid volume.

To evaluate the association of PiB SUVR with GDS, we performed multiple linear regression analysis entering age and sex as covariates.

We conducted path analyses to evaluate whether white matter network change mediated the relationship between CSVD markers and GDS after controlling for age, and sex. In order to investigate the independent effects of CSVD on depressive symptoms regardless of amyloid burden, we controlled for PiB SUVR. Path analysis was used to simultaneously consider the direct, indirect, and total effects of predictors on outcomes through mediators. Each network parameter was used as a mediator (path analysis A to E). We used the bootstrapping method to verify the significance of the indirect effects. Amos Version 18.0 software (SPSS, Chicago, IL, USA) was used for all path analyses using maximum likelihood estimation.

RESULTS

Demographics

The characteristics of our study participants are shown in Table 1. The mean age of participants was 72.1 years (SD 8.1) and 56.5% were female. 55.2% of participants showed amyloid positivity. The mean GDS of the SVaD group was 16.0 (SD 8.3), which was higher than that of the aMCI and AD groups. The mean WMH volume was 22.9 ml (SD 22.5), and mean number of lacune was 6.7 (SD 12.1).

Table 1

Demographics and clinical and neuroimaging features of participants

	Total	aMCI	AD	svMCI	SVaD	p
	(N = 228)	(N = 43)	(N = 62)	(N = 58)	(N = 65)
Demographics
Age, y	72.0 (8.1)	70.1 (7.9)	69.6 (9.6)	74.1 (6.6)	73.7 (7.2)	0.002
Sex, female, n (%)	128 (56.1)	20 (46.5)	39 (62.9)	33 (56.9)	36 (55.4)	0.424
Education, y	10.2 (5.3)	12.4 (4.7)	10.5 (5.7)	9.9 (5.3)	8.7 (5.1)	0.004
Cardiovascular risk factors, n (%)
Hypertension	144 (61.8)	16 (37.2)	33 (53.2)	42 (72.4)	50 (76.9)	<0.001
Diabetes mellitus	46 (20.2)	5 (10.9)	9 (14.5)	16 (27.6)	16 (24.6)	0.114
Hyperlipidemia	67 (29.4)	10 (23.3)	15 (24.2)	18 (31.0)	24 (36.9)	0.329
Cardiac disease	33 (14.5)	7 (16.3)	5 (8.1)	15 (25.9)	6 (9.2)	0.021
APOE genotype, n (%)^†
ɛ2 allele carrier	18/221 (8.1)	3/41 (7.3)	3/60 (5.0)	7/58 (12.1)	5/62 (8.1)	0.567
ɛ4 allele carrier	79/221 (35.7)	16/41 (39.0)	32/60 (53.3)	13/58 (22.4)	18/62 (29.0)	0.003
Small vessel disease MRI markers
WMH volume, ml	22.9 (22.5)	3.2 (3.2)	4.6 (6.0)	34.1 (17.9)	43.3 (18.6)	<0.001
Lacunes, n	6.9 (12.1)	0.6 (2.2)	0.3 (0.8)	7.6 (8.5)	16.6 (17.0)	<0.001
Amyloid PET markers
PiB SUVR	1.8 (0.5)	1.8 (0.5)	2.2 (0.4)	1.5 (0.4)	1.5 (0.5)	<0.001
MMSE	22.8 (5.1)	26.1 (2.5)	19.3 (4.5)	26.4 (2.8)	20.6 (4.9)	<0.001
CDR-SB	3.8 (3.3)	1.6 (1.0)	5.0 (2.4)	1.4 (1.1)	6.3 (3.8)	<0.001
GDS	13.3 (7.5)	11.4 (6.8)	12.2 (7.1)	12.8 (6.7)	16.0 (8.3)	0.005
Network parameters
Shortest path length	1.22 (0.08)	1.18 (0.04)	1.20 (0.07)	1.22 (0.08)	1.27 (0.10)	<0.001
Global efficiency	0.88 (0.03)	0.90 (0.02)	0.90 (0.03)	0.88 (0.03)	0.86 (0.03)	<0.001
Clustering coefficient	3.96 (0.86)	3.48 (0.44)	3.54 (0.62)	4.02 (0.78)	4.64 (0.88)	<0.001
Transitivity	3.53 (0.76)	3.05 (0.36)	3.17 (0.52)	3.60 (0.74)	4.15 (0.74)	<0.001
Modularity	0.53 (0.51)	0.51 (0.34)	0.51 (0.45)	0.54 (0.05)	0.57 (0.04)	<0.001

aMCI, amnestic mild cognitive impairment; AD, Alzheimer’s disease; svMCI, subcortical vascular mild cognitive impairment; SVaD, subcortical vascular dementia; APOE, apolipoprotein E; MRI, magnetic resonance imaging; WMH, white matter hyperintensities; PET, positron emission tomography; PiB, Pittsburgh compound B; SUVR, standardized uptake value ratio; MMSE, Mini-Mental State Examination; CDR-SB, clinical dementia rating-sum of boxes; GDS, Geriatric Depression Scale; ^†APOE genotyping was performed in 221 out of 228 participants. Values are expressed as the means (standard deviations) or numbers (%). One-way analysis of variance and chi-square tests were used for testing statistical significance among groups.

Relationships of CSVD and amyloid imaging markers with GDS

WMH burden (log_e (WMH volume +1)) was associated with GDS (beta = 0.045, p = 0.040, Table 2). Number of lacunes was also associated with GDS (beta = 0.181, p < 0.001, Table 2). However, there was no correlation between PiB SUVR and GDS (Supplementary Table 1).

Table 2

Relationships between neuroimaging markers and GDS

	β	SE	p
Small vessel disease MRI markers
WMH burden, ln (WMH +1)	0.97	0.44	0.029
Lacunes, n	0.18	0.04	<0.001
Network parameters
Shortest path length	15.03	6.01	0.009
Global efficiency	–50.65	16.07	0.002
Clustering coefficient	1.50	0.61	0.014
Transitivity	1.94	0.68	0.005
Modularity	22.24	10.57	0.037

GDS, Geriatric Depression Scale; β, unstandardized beta coefficient; SE, standard error; WMH, white matter hyperintensities; ln, natural log PET, positron emission tomography; PiB, Pittsburgh compound B. Values shown are the results of multiple linear regression for GDS. Covariates were age, sex, and PiB SUVR.

Relationships of white matter network measures and cortical thickness with GDS

Higher GDS was associated with increased shortest path length and decreased global efficiency, indicating decreased network integration. Higher GDS was also associated with increased clustering coefficient, transitivity, and modularity, indicating increased network segregation (Table 2). There was no correlation between cortical thickness and GDS after controlling for age and sex (Supplementary Table 1).

Path analysis for GDS

The model with ln (WMH volume +1), lacune number, and global efficiency showed the best to fit to the data: p-value for chi-square test = 0.696, comparative fit index (CFI) >0.999, and root mean square error of approximation (RMSEA) <0.001.

Global efficiency completely mediated the relationship between WMH and GDS. Global efficiency also partially mediated the relationship between lacunes and GDS (Table 3, Fig. 2). Total, direct, and indirect effects of predictors (WMH volume, number of lacunes) on GDS through global efficiency are shown in Supplementary Table 2. The results for other network parameters are shown in Supplementary Table 3.

Table 3

Effects of predictors (WMH volume, number of lacunes) on GDS through global efficiency

	Global efficiency			GDS
	β	SE	p	β	SE	p
WMH burden, ln (WMH +1)	–0.435	0.001	<0.001
Lacune, n	–0.182	0.000	0.004	0.209	0.043	0.002
Global efficiency				–0.166	18.540	0.016

WMH, white matter hyperintensities; ln, natural log; GDS, Geriatric Depression Scale; β, standardized beta coefficient; SE, standard error.

Fig. 2.

Schematic diagram of the path analyses for GDS. Natural log (WMH volume +1) and number of lacunes were entered as exposures. Decreased global efficiency completely mediated the relationship between WMH burden and GDS. The number of lacunes were associated with GDS and partially mediated by global efficiency. Numbers on the paths are standardized coefficients that were statistically significant.

DISCUSSION

In this study, we investigated the pathomechanism of LLD in cognitively impaired patients with varying degree of amyloid and vascular burdens. Our major findings were as follows: 1) CSVD burden, but not amyloid burden, was associated with increased GDS; 2) changes in white matter networks, but not cortical thinning, were related to GDS in the direction of decreasing integration and increasing segregation; 3) global efficiency completely mediated the relationship between WMH burden and GDS and partially mediated the relationship between number of lacunes and GDS. Therefore, our findings provide insight to better understand how CSVD burdens contribute to depression in cognitively impaired patients with varying degrees of amyloid and vascular burdens.

We found that CSVD markers, including WMH volume and number of lacunes, were associated with increased GDS. Our findings are consistent with previous studies showing that increased amount of WMH predicted the presence of depression [39 –42]. Recently, increased lacunes were also found to be associated with depressive symptoms [43]. However, findings regarding the possible association between amyloid burden and GDS have been inconsistent among studies. Recent studies have shown that presence of depressive symptoms is associated with increased amyloid burden in cognitively normal elderly, suggesting that depression may be a prodromal symptom of AD [9, 10]. However, a population-based postmortem study of people without dementia did not show any correlation between amyloid and depression [44]. Taken together, these findings suggest that depressive symptoms may be more strongly associated with CSVD burden than with amyloid burden in patients having various degrees of cognitive impairment.

Our major finding was that increased GDS was related to decreased network integration and increased network segregation, which is a characterization of what is called “regular” topology. Regular topology has increased short-range connections but decreased long-range connections, eventually resulting in inefficient organization [35]. Our finding is consistent with previous studies showing that patients with depression had decreased integration and relatively preserved or increased segregation [4, 45]. A previous study from our group also showed that disrupted networks related to CSVD markers had the characterization of regular topology. Our findings therefore suggest that, in relation to depressive symptoms, the network changes inefficiently in the direction of regular topology.

Another noteworthy finding in this study is that the global efficiency mediated the relationships between CSVD markers and GDS, suggesting that disrupted white matter network is important for the pathobiology of vascular depression. That is, ischemia in white matter areas causes demyelination and axonal loss of long projection fibers, which leads to reduced global efficiency of signal propagation [46, 47], eventually resulting in development of depression. Our suggestion appears to be supported by our other finding that cortical thickness was not correlated with GDS. Interestingly, global efficiency completely mediated the relationships between WMH burden and GDS and partially mediated the relationship between number of lacunes and GDS. It is still unclear why WMH burden and lacune number showed different paths for depression score through network changes. Further studies are needed to investigate the pathobiology of these different paths.

The pathologic burden of CSVD might cause cognitive decline and it is difficult to rule out the possibility that GDS was higher in patients with severe cognitive decline. However, in our study, the relationship between cognitive decline and depression was not significant when we examined the relationship between MMSE and GDS while controlling for age, sex, education, and PiB SUVR (β= –0.07; 95% CI, –0.15 to 0.01; p = 0.103). Therefore, CSVD burden seems to affect GDS directly rather than indirectly through cognitive decline.

Our study has some limitations. First, we estimated depressive symptom using GDS, which is a self-report questionnaire, so scores may be overestimated or underestimated. Second, because this study was cross-sectional in design, it can only indicate associations, and further longitudinal studies are needed to show causality. Third, we could not consider the effects of other pathologies, including other AD (soluble Aβ and neurofibrillary tangles), microinfarcts, or possible combined degenerative dementia (dementia with Lewy bodies and frontotemporal dementia) pathologies, which might also be associated with depression. Finally, we used the GDS value as an outcome indicating depressive symptoms rather than clinically diagnosed depression. In this study, we did not include patients with major depression diagnosed using the DSM-IV.

In conclusion, we investigated the relationships among CSVD, white matter network, and depression using multimodal analyses. We found that global efficiency mediated the relationships between CSVD markers and GDS. Our findings therefore provide insight to better understand how CSVD burdens contribute to depression in cognitively impaired patients having varying degrees of amyloid and vascular burdens.

Footnotes

ACKNOWLEDGMENTS

This research was supported by the Brain Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (2016M3C7A1913844) and the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIP) (No. NRF-2017R1A2B2005081).

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Krishnan

, Hays

, Tupler

, George

, Blazer

(1995) Clinical and phenomenological comparisons of late-onset and early-onset depression. Am J Psychiatry 152, 785–788.

Krishnan

, Taylor

, McQuoid

, MacFall

, Payne

, Provenzale

, Steffens

(2004) Clinical characteristics of magnetic resonance imaging-defined subcortical ischemic depression. Biol Psychiatry 55, 390–397.

Alexopoulos

, Meyers

, Young

, Kakuma

, Silbersweig

, Charlson

(1997) Clinically defined vascular depression. Am J Psychiatry 154, 562–565.

Ajilore

, Lamar

, Leow

, Zhang

, Yang

, Kumar

(2014) Graph theory analysis of cortical-subcortical networks in late-life depression. Am J Geriatr Psychiatry 22, 195–206.

Krishnan

, Hays

, Blazer

(1997) MRI-defined vascular depression. Am J Psychiatry 154, 497–501.

Alexopoulos

, Meyers

, Young

, Campbell

, Silbersweig

, Charlson

(1997) ‘Vascular depression’ hypothesis. Arch Gen Psychiat 54, 915–922.

Taylor

, Aizenstein

, Alexopoulos

(2013) The vascular depression hypothesis: Mechanisms linking vascular disease with depression. Mol Psychiatry 18, 963–974.

Aizenstein

, Baskys

, Boldrini

, Butters

, Diniz

, Jaiswal

, Jellinger

, Kruglov

, Meshandin

, Mijajlovic

, Niklewski

, Pospos

, Raju

, Richter

, Steffens

, Taylor

, Tene

(2016) Vascular depression consensus report – a critical update. BMC Med 14, 161.

Yasuno

, Kazui

, Morita

, Kajimoto

, Ihara

, Taguchi

, Yamamoto

, Matsuoka

, Kosaka

, Kudo

, Iida

, Kishimoto

, Nagatsuka

(2016) High amyloid-beta deposition related to depressive symptoms in older individuals with normal cognition: A pilot study. Int J Geriatr Psychiatry 31, 920–928.

10.

Mahgoub

, Alexopoulos

(2016) Amyloid hypothesis: Is there a role for antiamyloid treatment in late-life depression?. Am J Geriatr Psychiatry 24, 239–247.

11.

Diniz

, Butters

, Albert

, Dew

, Reynolds

CF 3rd

(2013) Late-life depression and risk of vascular dementia and Alzheimer’s disease: Systematic review and meta-analysis of community-based cohort studies. Br J Psychiatry 202, 329–335.

12.

Petersen

, Smith

, Waring

, Ivnik

, Tangalos

, Kokmen

(1999) Mild cognitive impairment: Clinical characterization and outcome. Arch Neurol 56, 303–308.

13.

Seo

, Cho

, Park

, Chin

, Na

(2009) Subcortical vascular versus amnestic mild cognitive impairment: Comparison of cerebral glucose metabolism. J Neuroimaging 19, 213–219.

14.

Ahn

, Chin

, Park

, Lee

, Suh

, Seo

, Na

(2010) Seoul Neuropsychological Screening Battery-dementia version (SNSB-D): A useful tool for assessing and monitoring cognitive impairments in dementia patients. J Korean Med Sci 25, 1071–1076.

15.

, Kim

, Lee

, Ko

, Kwon

, Jo

, Kim

(2004) A study on the reliability and validity of Seoul-Activities of Daily Living (S-ADL). J Korean Geriatr Soc 8, 206–214.

16.

McKhann

, Drachman

, Folstein

, Katzman

, Price

, Stadlan

(1984) Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease. Neurology 34, 939–944.

17.

Erkinjuntti

, Inzitari

, Pantoni

, Wallin

, Scheltens

, Rockwood

, Roman

, Chui

, Desmond

(2000) Research criteria for subcortical vascular dementia in clinical trials. J Neural Transm Suppl 59, 23–30.

18.

Fazekas

, Kleinert

, Offenbacher

, Schmidt

, Kleinert

, Payer

, Radner

, Lechner

(1993) Pathologic correlates of incidental MRI white matter signal hyperintensities. Neurology 43, 1683–1689.

19.

Ahn

, Seo

, Chin

, Suh

, Lee

, Kim

, Im

, Lee

, Heilman

, Na

(2011) The cortical neuroanatomy of neuropsychological deficits in mild cognitive impairment and Alzheimer’s disease: A surface-based morphometric analysis. Neuropsychologia 49, 3931–3945.

20.

Kim

, Im

, Kwon

, Lee

, Ye

, Kim

, Cho

, Choe

, Lee

, Kim

, Lee

, Na

, Seo

(2015) Effects of amyloid and small vessel disease on white matter network disruption. J Alzheimers Dis 44, 963–975.

21.

Sporns

, Tononi

, Kötter

(2005) The human connectome: A structural description of the human brain. PLoS Comput Biol 1, e42.

22.

Banerjee

, Jang

, Kim

, Lee

, Im

, Kwon

, Lee

, Na

, Seo

, Werring

(2018) Total MRI small vessel disease burden correlates with cognitive performance, cortical atrophy, and network measures in a memory clinic population. J Alzheimers Dis 63, 1485–1497.

23.

Kim

, Im

, Kwon

, Lee

, Kim

, Jung

, Cho

, Ye

, Noh

, Kim

, Ko

, Kim

, Choe

, Lee

, Kim

, Lee

, Ewers

, Weiner

, Na

, Seo

(2015) Clinical effect of white matter network disruption related to amyloid and small vessel disease. Neurology 85, 63–70.

24.

Kim

, Yang

, Kwon

, Kim

, Lee

, Chun

, Kim

, Jung

, Chin

, Kim

, Woo

, Choe

, Lee

, Kim

, Lee

, Weiner

, Na

, Seo

(2016) Relative impact of amyloid-beta, lacunes, and downstream imaging markers on cognitive trajectories. Brain 139, 2516–2527.

25.

Jeon

, Yoon

, Park

, Seo

, Kim

, Na

, Lee

(2011) Fully automated pipeline for quantification and localization of white matter hyperintensity in brain magnetic resonance image. Int J Imag Syst Tech 21, 193–200.

26.

Wardlaw

, Smith

, Biessels

, Cordonnier

, Fazekas

, Frayne

, Lindley

, O’Brien

, Barkhof

, Benavente

, Black

, Brayne

, Breteler

, Chabriat

, Decarli

, de

Leeuw FE

, Doubal

, Duering

, Fox

, Greenberg

, Hachinski

, Kilimann

, Mok

, Oostenbrugge

, Pantoni

, Speck

, Stephan

, Teipel

, Viswanathan

, Werring

, Chen

, Smith

, van

Buchem M

, Norrving

, Gorelick

, Dichgans

; STandards for ReportIng Vascular changes on nEuroimaging (STRIVE v1) (2013) Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 12, 822–838.

27.

Collins

, Neelin

, Peters

, Evans

(1994) Automatic 3D intersubject registration of MR volumetric data in standardized Talairach space. J Comput Assist Tomogr 18, 192–205.

28.

Tzourio-Mazoyer

, Landeau

, Papathanassiou

, Crivello

, Etard

, Delcroix

, Mazoyer

, Joliot

(2002) Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 15, 273–289.

29.

Mori

, Crain

, Chacko

, van

Zijl PC

(1999) Three-dimensional tracking of axonal projections in the brain by magnetic resonance imaging. Ann Neurol 45, 265–269.

30.

Hagmann

, Sporns

, Madan

, Cammoun

, Pienaar

, Wedeen

, Meuli

, Thiran

, Grant

(2010) White matter maturation reshapes structural connectivity in the late developing human brain. Proc Natl Acad Sci U S A 107, 19067–19072.

31.

Reijmer

, Freeze

, Leemans

, Biessels

, Utrecht

Vascular Cognitive Impairment Study Group

(2013) The effect of lacunar infarcts on white matter tract integrity. Stroke 44, 2019–2021.

32.

Shu

, Liu

, Li

, Duan

, Wang

, Yu

, Dong

, Ye

, He

(2011) Diffusion tensor tractography reveals disrupted topological efficiency in white matter structural networks in multiple sclerosis. Cereb Cortex 21, 2565–2577.

33.

Rubinov

, Sporns

(2010) Complex network measures of brain connectivity: Uses and interpretations. Neuroimage 52, 1059–1069.

34.

Watts

, Strogatz

(1998) Collective dynamics of ‘small-world’ networks. Nature 393, 440–442.

35.

Latora

, Marchiori

(2001) Efficient behavior of small-world networks. Phys Rev Lett 87, 198701.

36.

Onnela

, Saramaki

, Kertesz

, Kaski

(2005) Intensity and coherence of motifs in weighted complex networks. Phys Rev E Stat Nonlin Soft Matter Phys 71, 065103.

37.

Blondel

, Guillaume

J-L

, Lambiotte

, Lefebvre

(2008) Fast unfolding of communities in large networks. J Stat Mech 2008 P10008.

38.

Kang

, Na

(2003) Seoul Neuropsychological Screening Battery. Human Brain Research & Consulting, Incheon.

39.

Taylor

, MacFall

, Payne

, McQuoid

, Steffens

, Provenzale

, Krishnan

(2005) Greater MRI lesion volumes in elderly depressed subjects than in control subjects. Psychiatry Res 139, 1–7.

40.

Greenwald

, Kramer-Ginsberg

, Krishnan

, Ashtari

, Aupperle

, Patel

(1996) MRI signal hyperintensities in geriatric depression. Am J Psychiatry 153, 1212–1215.

41.

Kumar

, Bilker

, Lavretsky

, Gottlieb

(2000) Volumetric asymmetries in late-onset mood disorders: An attenuation of frontal asymmetry with depression severity. Psychiatry Res 100, 41–47.

42.

Herrmann

, Le

Masurier M

, Ebmeier

(2008) White matter hyperintensities in late life depression: A systematic review. J Neurol Neurosurg Psychiatry 79, 619–624.

43.

Santos

, Gold

, Kovari

, Herrmann

, Bozikas

, Bouras

, Giannakopoulos

(2009) Differential impact of lacunes and microvascular lesions on poststroke depression. Stroke 40, 3557–3562.

44.

Tsopelas

, Stewart

, Savva

, Brayne

, Ince

, Thomas

, Matthews

, Medical

Research Council Cognitive Function and Ageing Study

(2011) Neuropathological correlates of late-life depression in older people. Br J Psychiatry 198, 109–114.

45.

Bai

, Shu

, Yuan

, Shi

, Yu

, Wu

, Wang

, Xia

, He

, Zhang

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

46.

Franklin

, Ffrench-Constant

(2008) Remyelination in the CNS: From biology to therapy. Nat Rev Neurosci 9, 839–855.

47.

Iadecola

(2013) The pathobiology of vascular dementia. Neuron 80, 844–866.