Cortical atrophy, reduced integrity of white matter and cognitive impairment in subcortical vascular dementia of Binswanger type

Authors

  • Won-Beom Jung MS,

    1. Department of Biomedical Engineering and FIRST, Inje University, Gimhae, Korea
    Search for more papers by this author
  • Chi-Woong Mun PhD,

    1. Department of Biomedical Engineering and FIRST, Inje University, Gimhae, Korea
    Search for more papers by this author
  • Young-Hoon Kim MD,

    1. Department of Psychiatry, Medical School, Haeundae Paik Hospital, Inje University, Busan, Korea
    Search for more papers by this author
  • Je Min Park MD, PhD,

    1. Department of Psychiatry, Pusan National University Hospital, Busan, Korea
    2. Medical Research Institute, Pusan National University Hospital, Busan, Korea
    Search for more papers by this author
  • Byung Dae Lee MD, PhD,

    1. Department of Psychiatry, Pusan National University Hospital, Busan, Korea
    2. Medical Research Institute, Pusan National University Hospital, Busan, Korea
    Search for more papers by this author
  • Young Min Lee MD,

    Corresponding author
    1. Department of Psychiatry, Pusan National University Hospital, Busan, Korea
    2. Medical Research Institute, Pusan National University Hospital, Busan, Korea
    • Correspondence: Young Min Lee, MD, Department of Psychiatry, School of Medicine, Pusan National University, 305 Gudeok-Ro, Seo-Gu, Busan 602-739, Korea. Email: psyleekr@naver.com

    Search for more papers by this author
  • Eunsoo Moon MD,

    1. Department of Psychiatry, Pusan National University Hospital, Busan, Korea
    2. Medical Research Institute, Pusan National University Hospital, Busan, Korea
    Search for more papers by this author
  • Hee Jeong Jeong MD,

    1. Department of Psychiatry, Pusan National University Hospital, Busan, Korea
    2. Medical Research Institute, Pusan National University Hospital, Busan, Korea
    Search for more papers by this author
  • Young In Chung MD, PhD

    1. Department of Psychiatry, Pusan National University Yangsan Hospital, Yangsan, Korea
    Search for more papers by this author

Abstract

Aims

An association between white matter hyperintensities (WMH) and cognitive dysfunction has long been recognized. However, subjects with identically appearing WMH on magnetic resonance imaging present with a wide variance in cognitive function ranging from normal cognition to dementia. The aim of this study was to compare cortical atrophy and integrity of white matter of patients with subcortical vascular dementia of Binswanger type (SVaD-BT) with those of the normal cognition group with WMH (ncWMH).

Methods

Eleven patients with SVaD-BT and 11 age-, sex-, education- and grade of WMH-matched ncWMH underwent magnetic resonance imaging, including 3-D volumetric images for cortical atrophy and diffusion tensor imaging for integrity of white matter.

Results

Compared to ncWMH, SVaD-BT patients showed cortical atrophies in frontal (i.e. frontal pole, precentral gyrus and frontal medial cortex) and occipital areas (i.e. lingual gyrus) followed by atrophies in temporal (i.e. fusiform cortex and middle temporal gyrus) areas. Along with cortical atrophies, reduced integrity with low fractional anisotropy and high mean diffusivity values in genu and splenium of the corpus callosum were detected in SVaD-BT patients.

Conclusions

Our findings suggest that cognitive decline from ncWMH to SVaD-BT may be associated with cortical atrophy and reduced integrity of white matter.

White matter hyperintensities (WMH) are a frequent finding on magnetic resonance imaging (MRI) scans of elderly people. It is known that WMH are associated with cognitive decline and neurobehavioral symptoms, such as apathy, abulia, agitation, disinhibition, reduced mental speed, impaired executive functions, and relatively mild memory dysfunction.[1, 2] Cognitive impairments in WMH are probably related to ischemic interruption of frontal subcortical circuits or disruption of cholinergic pathways that traverse the subcortical white matter (WM).[2, 3]

However, subjects with identically appearing WMH on MRI present with a wide variance in cognitive function ranging from no complaints to dementia.[4] It is now appreciated that these lesions are prevalent both in normal aging and in aging associated with cognitive decline. It means that there are hidden factors that determine whether identically appearing WMH on MRI lead to cognitive decline in one person, while leaving others unaffected.

One of the hidden factors could be neurodegenerative processes that are triggered by WMH. Previous studies have shown that neuronal degeneration occurs when an axon in the brain is damaged.[5, 6] This would eventually result in atrophy in cortical areas that are connected via damaged tracts.

Another explanation for the clinical variety due to WMH could be changes of microstructural integrity that are invisible on conventional MRI but are nevertheless expected to contribute substantially to clinical symptoms.[7] Unfortunately, conventional MRI commonly used in clinical practice are insufficiently sensitive and specific to detect all the WM changes related to small vessel diseases. These limitations of conventional MRI can potentially be overcome with the use of diffusion tensor imaging (DTI), which is a sensitive tool for detecting microstructural changes in the normally appearing WM on conventional MRI.[8]

The aim of this study was to compare the regional distribution of cortical atrophy of subcortical vascular dementia of Binswanger type (SVaD-BT) with that of the normal cognition group with WMH (ncWMH) using voxel-based morphometry (VBM). Another aim was to compare the changes of microstructural WM integrity in SVaD-BT with that of ncWMH using DTI.

Methods

Subjects

This study involved 11 patients with SVaD-BT and 11 age-, sex- and education-matched ncWMH. All subjects were recruited from the memory-impairment clinics of Pusan National University Hospital in Korea from November 2010 to March 2012.

Patients with SVaD-BT met the criteria for VaD described by the DSM-IV and also fulfilled the imaging criteria proposed by Erkinjuntti et al.[9] We applied the following exclusion criteria to all subjects: (i) other neurodegenerative or psychiatric diseases; (ii) intracranial space-occupying lesion; (iii) prominent visual or hearing impairment; (iv) aphasia or other language barrier; (v) MRI contraindications or known claustrophobia; (vi) active substance abuse disorders; and (vii) severe systemic disease. Present or past use of medication to treat cognition and infarction was not included in the exclusion criteria. The inclusion criteria for ncWMH were as follows: (i) subjective cognitive complaints but no abnormality (within −1.5 SD of age- and education-adjusted norms) on standardized neuropsychological tests; (ii) normal activities of daily living (ADL) as judged by both an interview with a clinician and the standardized ADL scale; and (iii) significant small vessel ischemic changes without territory infarction on MRI. The presence of significant small vessel ischemic changes was defined as WMH on fluid-attenuated inversion recovery (FLAIR) imaging that fulfilled the following criteria: (i) periventricular WMH (caps or rim) longer than 10 mm; and (ii) deep WMH consistent with extensive WM lesion or diffusely confluent lesion ≥25 mm in maximum diameter. These imaging criteria indicate that our patients had ischemia significant enough to meet at least grade 3 of Fazekas' ischemia criteria.[10]

We used the Korean version of the Consortium to Establish a Registry for Alzheimer's Disease (CERAD-K) neuropsychological battery to examine the functional capacity of several cognitive domains:[11] (i) memory (word list recall, constructional recall); (ii) language (the Korean version of the Boston Naming Test [K-BNT]); and (iii) visuospatial function (constructional apraxia). We also used the Korean version of the Frontal Assessment Battery (FAB-K)[12] to examine the frontal or executive function: the FAB-K was known as a valid and reliable instrument for evaluating frontal lobe function in the elderly.

Written informed consent was obtained from all subjects, and this study was approved by the Pusan National University Hospital Institutional Review Board. The demographic and clinical features of the subjects are summarized in Table 1.

Table 1. Demographic information of the subjects
 SVaD-BT (n = 11)ncWMH (n = 11)P-value
  1. Data are n (%) or means ± SD.
  2. CERAD-K, Korean version of Consortium to Establish a Registry for Alzheimer's disease; DM, diabetes mellitus; FAB, Frontal Assessment Battery; MMSE, Mini Mental State Examination; ncWMH, normal cognition group with white matter hyperintensity; SIADL, Seoul-Instrumental Activities of Daily Living; SVaD-BT, subcortical vascular dementia of Binswanger type.
Age, years73 ± 4.2775 ± 1.410.834
Sex, female, n (%)991.000
Education, years7 ± 4.168 ± 6.360.833
Hypertension, n (%)5 (45.5)6 (54.5)0.670
DM, n (%)4 (36.4)2 (18.2)0.635
Hyperlipidemia, n (%)3 (27.3)2 (18.2)1.000
Any APOE e4 allele, n (%)2 (18.2)3 (27.3)1.000
MMSE16.82 ± 5.3124.73 ± 4.410.003
CERAD-K   
Boston naming test6.73 ± 2.949.45 ± 4.520.118
Constructional apraxia6.09 ± 2.669.63 ± 1.570.001
Word list recall1.27 ± 1.492.64 ± 2.380.152
Constructional recall0.72 ± 1.013.18 ± 2.710.018
FAB8.55 ± 3.1411.91 ± 2.510.015
SIADL22.73 ± 13.997.0 ± 4.750.005

MRI data acquisition

All subjects underwent MR scans of T1-weighted images (T1WI) and DTI on a Siemens (Erlangen, Germany) Trio TIM 3T scanner. T1WI were acquired using a 3-D magnetization-prepared rapid gradient echo (3D MPRAGE) sequence with the following parameters: repetition time (TR) = 1800 ms, echo time (TE) = 2.07 ms, inversion time (TI) = 900 ms, flip angle = 12°, acquisition matrix = 256 × 256, field of view (FOV) = 250 × 250 mm2, slice thickness = 1 mm, and total number of slices = 256. DTI were acquired with the following echo planar acquisition parameters: diffusion-weighted gradients applied in 30 non-linear directions, number of average = 2, TR = 6200 ms, TE = 85 ms, flip angle = 90°, acquisition matrix = 128 × 128, FOV = 230 × 230 mm2, slice thickness = 3 mm, and b value = 600 s/mm2. We ensured the same slice orientation paralleled with the anterior commissure and posterior commissure line in all image acquisitions.

MRI data analysis

Cortical volume atrophy: VBM analysis

The VBM 8 toolbox (http://dbm.neuro.uni-jena.de/vbm), which is incorporated in SPM8 (http://www.fil.ion.ucl.ac.uk/spm),[13] was used to perform the analysis of brain structural imaging. In this process, all images were spatially normalized using combinations of linear transform and non-linear registration to the standard Montreal Neurological Institute (MNI) template and segmented into probabilistic gray matter (GM), WM and cerebrospinal fluid (CSF).[14] Because WMH areas were represented in similar intensity ranges with GM, segmentation processing was implemented by applying an adaptive maximum a posteriori technique. This technique relied solely on intensity differences without requiring the a priori information for tissue class probabilities and by additionally adapting the partial volume estimation with hidden Markov random field approach.[15-17] Additionally, FreeSurfer 5.1 (http://surfer.nmr.mgh.harvard.edu/)[18, 19] was applied to confirm the WMH and cerebral cortex in T1WI. FreeSurfer is a highly automatic tool for brain structural segmentation comparable in accuracy to manual assessment. Briefly, the images are processed with motion correction, a non-uniform intensity correction, removal of non-brain tissues, Talairach registration, segmentation of cortical and subcortical GM based on manually generated probabilistic atlas and construction of surface models with boundary tessellation and topological correction. The more detailed procedures with Freesurfer have been described in previous studies.[18, 19] WMH areas roughly are labeled as hypointense areas in T1WI in subcortical segmentation procedures. Although this performance was lower than that from manual or supervised methods, its usefulness was reported in previous studies.[20, 21] To perform the statistical analysis only in the cerebral cortex, the intersection between probabilistic GM from VBM8 and cortical areas from Freesurfer were estimated. Then, segmented GM images were modulated to compensate the volumetric effects of expansion or shrinking employed in spatial normalization by multiplying the voxel intensity with the Jacobian determinants. This reflected the parameters for fitting a voxel in native space to a corresponding voxel in MNI space. The processing for identification of cortical areas is shown in Figure 1. The modulated images were then smoothed with an 8-mm full-width half maximum (FWHM) isotropic Gaussian kernel. Finally, statistical analysis was designed using an independent two-sample t-test with general linear model (GLM). Age, sex, education and intracranial volume were used as covariates to control these confounding factors. A statistically significant level was considered at false discovery rate (FDR) corrected P-values (q < 0.05) with the extent threshold of contiguous 500 voxels.

Figure 1.

Identification of cortical areas. (a) Whole brain image in native space. (b) White matter hyperintense (blue) and cortical areas (red) from Freesurfer overlaid on (a). (c) Probabilistic gray matter from SPM. (d) Intersection gray matter between (c) and cortical areas from Freesurfer. (e) Normalized into Montreal Neurological Institute (MNI) space from (c). (f) Normalized into MNI space from (d).

Microstructural integrity in white matter: TBSS analysis

DTI were entirely processed with the FMRIB Software Library (FSL, http://www.fmrib.ox.ac.uk/fsl) package.[22] Image distortions induced by the effects of head movement and eddy currents were corrected by applying an alignment of diffusion-weighted images to non-diffusion weighted images. After stripping the skull using brain extraction tool (BET), diffusion tensor was calculated at each voxel and then fractional anisotropy (FA), mean diffusivity (MD (λ1 + λ2 + λ3)/3), axial diffusivity (DA, λ1) and radial diffusivity (RD (λ2 + λ3)/2) images were generated using the fMRIB diffusion toolbox (FDT).[22, 23] Voxel-wise statistical analysis of these DTI parameters was performed using TBSS 1.2.[24, 25] First, all individual FA images were non-linearly aligned to the FA template in 1 × 1 × 1 mm3 MNI space using FMRIB's non-linear registration tool (FNIRT).[22] Second, aligned images were averaged and thinned to identify the mean FA skeleton represented as the center of common WM tracts.[24] Third, mean FA skeleton was restricted with the thresholds 0.2 to exclude the voxels of GM, CSF in skeleton. All of the subject's aligned FA images were projected onto the mean skeleton tracks and the resulting data fed into voxel-wise across subject statistics. The registration and projection vectors derived from DTI-FA processing were applied to other DTI parameters. Finally, voxel-wise differences of DTI parameters were statistically analyzed by independent two-sample t-test with the GLM design controlling the effects of age, sex and education as covariates, setting the number of permutations at 5000 based on the threshold-free cluster enhancement (TFCE) and taking into account results to be significant at P < 0.05 with correction for multiple comparisons.[26]

Automatic region of interest analysis in cerebral cortex and corpus callosum

Following voxel-wise comparisons, we performed region of interest (ROI) analysis based on the VBM and TBSS results. Cortical volumes were estimated from segmentation in Freesurfer and WM integrities based on DTI parameters were measured using skeleton deprojection into the subject's native space. Common areas showing statistical differences between the two groups in TBSS analysis were identified in reference to JHU ICBM-DTI-81 WM atlas.[27] This atlas in standard space was labeled into sub-anatomical WM ROI and then these were transformed to all of the subject's native spaces using the inverse normalization with both normalization matrix to standard space and skeleton projection vectors applied in TBSS analysis. Back-transformed atlas masks were multiplied by the FA images and acquired the averaging FA values within the ROI. The same processing was conducted for other DTI parameters to evaluate the original non-FA values. After assessing the normal distribution by Kolmogorov–Smirnov test, cortical volumes and mean DTI ROI values were compared using the Mann–Whitney U-test with significance level of P < 0.05. We also performed the correlation analysis between the severity of clinical features and brain ROI values defined by VBM or TBSS.

Results

Cortical gray matter volume: VBM analysis

Table 2 describes the anatomical areas, cluster size and T-scores for significantly different cortical volumes (FDR corrected q < 0.05 with k > 500). SVaD-BT compared with ncWMH showed the cortical atrophy in the following order on cluster sizes: frontal (i.e. frontal pole, precentral gyrus and frontal medial cortex) > occipital (i.e. lingual gyrus) > temporal (fusiform cortex and middle temporal gyrus) areas (Fig. 2). No supra-threshold clusters of increased areas were found in SVaD-BT with respect to ncWMH.

Table 2. Areas showing decreased cerebral gray matter volume with SVaD-BT compared to ncWMH on VBM analysis (FDR corrected q < 0.05 with k > 500)Thumbnail image of
Figure 2.

Voxel-based morphometry results showing the decreased cerebral volume in subcortical vascular dementia of Binswanger type compared to normal cognition group with white matter hyperintensities (false discovery rate corrected q < 0.05 with k > 500). The brain images are represented in Montreal Neurological Institute space with Z-coordinate. The color bar indicates the range of T-scores.

Microstructural integrity in white matter: TBSS analysis

Table 3 summarizes the results of DTI parameters within the maximum statistical thresholds. FA was lowered (P < 0.07) in the body of the corpus callosum (CC) and superior longitudinal fasciculus in SVaD-BT relative to ncWMH (Fig. 3a). MD was significantly high (P < 0.05) in the genu and body of the CC (Fig. 3b). Higher RD was noted (P < 0.05) in the genu of the CC and superior longitudinal fasciculus (Fig. 3c) but DA increased (P < 0.08) only in the superior corona radiate (Fig. 3d).

Table 3. Areas showing changed white matter integrity in SVaD-BT compared to ncWMH on TBSS analysis
Anatomical regionSidePeak α-valueMNI coordinates (mm)Cluster size (mm3)
xyz
  1. Peak α-value, 1–P-value.
  2. DA, axial diffusivity; FA, fractional anisotropy; I, inter hemispheric; L, left; MD, mean diffusivity; MNI, Montreal Neurological Institute; ncWMH, normal cognition group with white matter hyperintensity; R, Right; RD, radial diffusivity; Sup, superior; SVaD-BT, subcortical vascular dementia of Binswanger type.
FA decreased      
Body of corpus callosumR0.93391920769
Sup longitudinal fasciculusR0.94617649269
MD increased      
Genu of corpus callosumI0.95902016379
Body of corpus callosumR0.9545−126170
RD increased      
Genu of corpus callosumR0.969220161312
Sup longitudinal fasciculusR0.96716550245
DA increased      
Sup corona radiateR0.92727−222419
Figure 3.

Topology of TBSS analysis for significantly changed diffusion tensor imaging parameters in patients with subcortical vascular dementia of Binswanger type (SVaD-BT) compared to normal cognition group with white matter hyperintensities (ncWMH): (a) ncWMH > SVaD-BT (P < 0.07), (b) SVaD-BT > ncWMH (P < 0.05), (c) SVaD-BT > ncWMH (P < 0.05), (d) SVaD-BT > ncWMH (P < 0.08), corrected for family wise errors. The brain images are represented in Montreal Neurological Institute space with X, Y and Z coordinates. The color bar indicates the range of α-value (1-p). DA, axial diffusivity; FA, fractional anisotropy; MD, mean diffusivity; RD, radial diffusivity.

Automatic ROI analysis in cerebral cortex and corpus callosum

Figure 4a shows the cortical volumes in predefined areas between SVaD-BT and ncWMH. The overall cortical volumes of ncWMH were significantly larger than those of SVaD-BT, which was similar with VBM analysis, except for middle temporal gyrus. CC was commonly changed in the TBSS analysis between the two groups, although FA values were not significant (P < 0.07). Therefore, CC as ROI to analysis DTI parameters was classified into three areas of genu, body and splenium in this study. The acquired data and significance are presented in Figure 4b. The FA values of SVaD-BT in genu and splenium were significantly less than those of ncWMH. The MD and RD values were significantly high in all areas of CC and only in splenium for DA.

Figure 4.

Region of interest-based analysis in (a) cortical areas showing the volumetric differences in voxel-based morphometry and in (b) corpus callosum for diffusion tensor imaging parameters between normal cognition group with white matter hyperintensities (ncWMH) and patients with subcortical vascular dementia of Binswanger type (SVaD-BT). The asterisks (*,**) represent the significant differences (P < 0.05, P < 0.01). FC, fusiform cortex; FMC, frontal medial cortex; FPC, frontal pole cortex; LG, lingual gyrus; MTG, middle temporal gyrus; PCG, precentral gyrus. image, ncWMH_Left; image, SvaD-BT_Left; image, ncWMH_Right; image, SvaD-BT_Right; image, ncWMH_fractional anisotropy (FA); image, SvaD-BT_FA; image, ncWMH_mean diffusivity (MD); image, SvaD-BT_MD; image, ncWMH_axial diffusivity (DA); image, SvaD-BT_DA; image, ncWMH_radial diffusivity (RD); image, SvaD-BT_RD.

In the correlation analysis, as shown in Figure 5, inter-clinical features of the subjects were positively correlated with each other except for the Seoul-Instrumental Activities of Daily Living; in particular, the relationships with Mini Mental State Examination (MMSE) scores were relatively high (R > 0.6). Most of the correlations between clinical features and cortical volumes/inter-cortical volumes were positive. Fusiform cortex and lingual gyrus among the cortical areas showed high degrees (R > 0.6) of positive correlation with MMSE, BNT, constructional apraxia and FAB. On the other hand, DTI parameters, with exception of the FA values, displayed roughly negative correlation with clinical features and cortical volumes.

Figure 5.

Correlation analysis between neuropsychological tests and brain abnormalities derived from radial diffusivity and diffusion tensor imaging analysis. (a) Matrix of correlation coefficient. (b) Corresponding P-values. The color bar indicates the range of (a) R and (b) P-value. BNT, Boston naming test; CA, constructional apraxia; CR, constructional recall; DA, axial diffusivity; FA, fractional anisotropy; FAB, Frontal Assessment Battery; FC, fusiform cortex; FMC, frontal medial cortex; FPC, frontal pole cortex; LG, lingual gyrus; MD, mean diffusivity; MTG, middle temporal gyrus; PCG, precentral gyrus; RD, radial diffusivity; SIADL, Seoul-Instrumental Activities of Daily Living; WR, word list recall.

Discussion

An association between WMH and cognitive dysfunction has long been recognized. However, recent studies and clinical practice showed that MRI-identified WMH lesions appear to be sufficient to cause mild forms of cognitive dysfunction,[28] but rarely cause dementia in the absence of other brain pathologies.[29, 30] This means that there are hidden factors that determine whether identically appearing WMH on MRI lead to dementia, while leaving others unaffected. In this study, we explored whether the regional distribution of cortical atrophy and the changes of microstructural WM integrity differ between ncWMH and SVaD-BT, which may help us understand how ncWHM evolves into SVaD-BT. To the best of our knowledge, this is the first study investigating the pathogenetic differences underlying WMH between ncWMH and SVaD-BT with the same severity of WMH using VBM or DTI.

We found that compared to ncWMH, cortical atrophy in SVaD-BT was predominantly distributed in prefrontal (i.e. frontal pole cortex, precentral gyrus and frontal medial cortex) areas and temporo-occipital (i.e. lingual gyrus, fusiform gyrus and middle temporal gyrus) areas. Cortical atrophy in SVaD-BT that was detected in this study was highly consistent with that of previous studies.[31, 32] In particular, the occipital region is the main origin of the cortico-cortical visual pathway. The cortico-cortical visual pathway has the dorsal (motion and spatial information; ‘where is it?’) and ventral (form and color information; ‘what is it?’) visual streams that are relayed by feedforward projections through different lobes en route to the prefrontal cortex.[33] The ventral visual stream involves cortico-cortical visual projections that proceed from the ventral visual cortex of the lingual gyrus to the temporal fusiform cortex, as well as to inferior and middle temporal gyri. Some of this information is directed to the amygdala for emotional coding, and to the temporal tip of the superior and middle temporal gyri for visual–auditory language and naming functions. Other inferior and anterior temporal feedforward projections pass via the uncinate fasciculus en route to the frontal orbital cortex and frontal medial cortex.[33] By connecting these temporal and prefrontal areas, the uncinate fasciculus: (i) may be a crucial component of the system that regulates emotional responses to attaching emotional valence to visual information; (ii) is likely to be an important component of the circuit underlying recognition memory; and (iii) is implicated in cognitive tasks that are inextricably linked with emotional associations.[34] Thus, cortical atrophy (i.e. lingual gyrus, fusiform cortex, middle temporal gyrus and frontal medical cortex) involving the ventral visual stream of cortico-cortical visual pathway in our patients with SVaD-BT might lead to maladaptive behavior, such as disinhibition, apathy and abulia.

The dorsal stream proceeds from the lingual gyrus to somatosensory association cortex and posterior cingulate gyrus of the parietal cortex, and then to the prefrontal cortex for the spatial guidance of saccadic eye movements, and to the upper dorsolateral prefrontal cortex for the spatial aspects of short-term memory and executive functions.[33] Cortical atrophy of lingual gyrus and prefrontal cortex involving the dorsal visual stream in our patients with SVaD-BT might be associated with visual-spatial perceptual abnormalities, working memory problems and executive dysfunction.

Another main objective of our study was to compare the changes of microstructural WM integrity between ncWMH and SVaD-BT using DTI. The discrepancies of results between TBSS and ROI analysis currently employed in imaging studies came from the methodological differences. TBSS is the voxel-by-voxel comparisons on WM skeleton-applied non-parametric permutation testing for multiple comparisons to control any false positives, whereas ROI analysis is dependent on single statistic value of DTI parameters by voxel averaged within ROI. In our results, we found lower DTI-FA, higher DTI-MD and higher DTI-RD values of genu and splenium of CC in SVaD-BT patients compared with ncWMH patients, which suggests a loss of the normal microstructural organization within the genu and splenium. Few studies have been conducted to compare the changes of microstructural integrity in WM areas between ncWMH and SVaD-BT using DTI. Only a few studies comparing SVaD-BT and normal controls without WMH have been conducted.[35, 36] Sugihara et al. reported that compared to normal controls, the DTI-FA values of CC (genu and splenium) in SVaD-BT were also significantly lower.[35] CC, the largest main commissure fiber tract, establishes inter-hemispheric connections in a topographically organized way.[37] It may be divided into three areas: anterior (incorporating the genu), intermediary (body), and posterior (incorporating the splenium). The genu connects the prefrontal areas, rostral cingulated region and supplementary motor area. The body connects the posterior portion of the frontal lobes and the parietal lobes. The splenium connects the superior temporal fiber rostrally and the inferotemporal and preoccipital area caudally.[37, 38] Many studies reported that CC is associated with perception, attention, memory, language, reasoning, self-awareness, and creativity.[39-41] Thus, the impairment of microstructural integrity of genu and splenium in our SVaD-BT patients may be related to impairment of high-level inter-hemispheric integration.

Previous studies have suggested that the underlying mechanism of cortical atrophy in SVaD-BT may be concomitant Alzheimer's disease (AD) pathology or secondary axonal and transsynaptic degeneration following subcortical injury.[42] In a study comparing cortical atrophy between SVaD-BT and AD, Seo et al.[31] reported that frontal atrophy was predominant in SVaD-BT patients whereas AD patients exhibited atrophy mainly in the temporo-parietal areas. These results mean that cortical thinning in SVaD-BT cannot be explained entirely by concomitant AD pathology. Their finding may support the secondary-degeneration hypothesis. The underlying mechanism of the impairment of microstructural integrity in CC may be concomitant AD pathology, directly hypoxic ischemic change or secondary axonal and transsynaptic degeneration following WMH. However, directly hypoxic ischemic changes rarely develop in the CC, probably due to the rich arterial supply of the CC. In addition, according to a neuropathologic study, CC thickness correlated with brain weight in AD, and with the severity of deep WM lesions in SVaD-BT.[38] These results suggest that CC lesions in SVaD-BT may be secondary to deep WM lesions. Taken together, it seems that both cortical atrophy and CC lesions in SVaD-BT are due to secondary axonal and transsynaptic degeneration following WMH.

The results in this study have several limitations to be interpreted with caution. First, this study was limited by a relatively small sample size. Second, due to small sample size, although FDR-corrected (q < 0.05) thresholds were applied to identify the cortical atrophy in the VBM, correction for family wise error was not applied, which suggests the type I error may be higher. Third, we did not evaluate the quantitative difference between SVaD-BT and ncWMH in severity of WMH as FLAIR MPRAGE were not acquired. Although our subjects had ischemia that was significant enough to meet at least grade 3 of Fazekas' ischemia criteria, there may be a significant difference between SVaD-BT and ncWMH in severity of WMH. Finally, although the subjects included in this study were identified with subcortical vascular damages of Binswanger's type, some of the patients with extensive WMH may have significant amyloid burdens. Also, some patients may have other pathologic conditions, such as dementia with Lewy bodies or frontotemporal dementia. Therefore, further studies with a larger sample size and calibration of statistical thresholds would be required to verify our results and reduce any potential bias.

In conclusion, this study shows widespread cortical atrophy, including lingual gyrus and reduced integrity of the genu and splenium of the CC, in SVaD-BT patients compared with ncWMH. This means that cognitive decline from ncWMH to SVaD-BT may be associated with cortical atrophy and reduced WM integrity. However, the precise underlying mechanisms of these changes in SVaD-BT are currently not fully understood. To investigate the roles of the various underlying causes (degenerative vs vascular causes) on SVaD-BT, future studies with longitudinal study design and a larger number of subjects must be conducted.

Acknowledgments

This study was supported by a grant from the Daeho Ethnic Psychiatry Research Fund (2011). Pusan National University Hospital and the authors have no conflicts of interest.

Ancillary