White Matter Hyperintensities, Grey Matter Atrophy, and Cognitive Decline in Neurodegenerative Diseases

TL;DR: In this article, different spatial patterns and relationships between WMHs and grey matter atrophy in normal aging, individuals with mild cognitive impairment (MCI), Alzheimer's dementia (AD), fronto-temporal dementia (FTD), and de novo Parkinson's disease (PD).

Abstract: Introduction White matter hyperintensities (WMHs) as seen on T2w and FLAIR scans represent small-vessel disease related changes in the brain. WMHs are associated with cognitive decline in the normal aging population in general and more specifically in patients with neurodegenerative diseases. In this study, we assessed the different spatial patterns and relationships between WMHs and grey matter (GM) atrophy in normal aging, individuals with mild cognitive impairment (MCI), Alzheimer’s dementia (AD), fronto-temporal dementia (FTD), and de novo Parkinson’s disease (PD). Methods Imaging and clinical data were obtained from 3 large multi-center databases: The Alzheimer’s Disease Neuroimaging Initiative (ADNI), the frontotemporal lobar degeneration neuroimaging initiative (NIFD), and the Parkinson’s Progression Markers Initiative (PPMI). WMHs and GM atrophy maps were measured in normal controls (N= 571), MCI (N= 577), AD (N= 222), FTD (N= 144), and PD (N= 363). WMHs were segmented using T1w and T2w/PD or FLAIR images and mapped onto 45 white matter tracts using the Yeh WM atlas. GM volume was estimated from the Jacobian determinant of the nonlinear deformation field required to map the subject’s MRI to a standard template. The CerebrA atlas was used to obtain volume estimates in 84 GM regions. Mixed effects models were used to compare WMH in different WM tracts and volume of multiple GM structures between patients and controls, assess the relationship between regional WMHs and GM loss for each disease, and investigate their impact on cognition. Results MCI, AD, and FTD patients had significantly higher WMH loads than the matched controls. There was no significant difference in WMHs between PD and controls. For each cohort, significant interactions between WMH load and GM atrophy were found for several regions and tracts, reflecting additional contribution of WMH burden to GM atrophy. While these associations were more relevant for insular and parieto-occipital regions in MCI and AD cohorts, WMH burden in FTD subjects had greater impact on frontal and basal ganglia atrophy. Finally, we found additional contribution of WMH burden to cognitive deficits in AD and FTD subjects compared with matched controls, whereas their impact on cognitive performance in MCI and PD were not significantly different from controls. Conclusions WMHs occur more extensively in MCI, AD, and FTD patients than age-matched normal controls. WMH burden on WM tracts also correlates with regional GM atrophy in regions anatomically and functionally related to those tracts, suggesting a potential involvement of WMHs in the neurodegenerative process.

Summary

Introduction

• These age-related WMHs are considered to be the most common MRI signs of cerebral small vessel disease and are generally due to chronic hypoperfusion and alterations in the blood brain barrier (McAleese et al., 2016).
• Few studies have investigated the relationship between the longitudinal changes in WMHs in different white matter tracts, neurodegenerative changes, and cognitive decline.
• They found significantly greater total load of WMHs in AD, but not PD or DLB.
• They did not investigate the relationships with measures of grey matter atrophy.

ADNI

• The longitudinal MRI data used in this study included T1w, T2w/proton density–weighted acquisitions from ADNI1 patients and T1w and FLAIR acquisitions from ADNI2/GO patients.
• The scanner information and image acquisition parameters have been previously described (Dadar et al., 2017a).
• The ADNI1, ADNI2 and ADNIgo studies acquired data from subjects on a yearly basis.

PPMI

• //www.ppmi-info.org) is a longitudinal multi-site clinical study of approximately 600 de novo PD patients and 200 age-matched healthy controls followed over the course of five years (Marek et al., 2011), also known as The PPMI (http.
• The study was approved by the institutional review board of all participating sites and written informed consent was obtained from all participants before inclusion in the study.

NIFD

• The frontotemporal lobar degeneration neuroimaging initiative is founded through the National Institute of Aging and started in 2010.
• The primary goals of FTLDNI are to identify neuroimaging modalities and methods of analysis for tracking frontotemporal lobar degeneration (FTLD) and to assess the value of imaging versus other biomarkers in diagnostic roles.
• The Principal Investigator of FTLDNI is Dr. Howard Rosen, MD at the University of California, San Francisco.
• The data is the result of collaborative efforts at three sites in North America.
• For up-todate information on participation and protocol, please visit: http://memory.ucsf.edu/research/studies/nifd.

WMHs

• All T1-weighted, T2-weighted, proton density (PD), and FLAIR MRI scans were preprocessed in 3 steps using their standardized pipeline: denoising (Manjón et al., 2010), intensity non-uniformity correction (Sled et al., 1998), and intensity normalization into the range 0–100.
• For each subject, the T2-weighted, PD, and FLAIR scans were then co-registered to the T1-weighted scan of the same visit using a 6-parameter rigid registration and a mutual information objective function (Collins et al., 1994; Dadar et al., 2018a).
• Using a previously validated fully automated WMH segmentation method and a library of manual segmentations based on 53 patients from ADNI1 and 46 patients from ADNI2/GO, the WMHs were automatically segmented for all longitudinal visits (Dadar et al., 2017b).
• The quality of the registrations and segmentations was visually assessed, and the results that did not pass this quality control were excluded (N=102 out of 5774 timepoints).

WM tracts and WMHs

• Using the atlas of the white matter tracts by Yeh et al. derived from diffusion MRI data of 842 young healthy individuals from the human connectome project (https://db.humanconnectome.org/) and labeled by a team of expert neuroanatomists based on tractography and neuroanatomical knowledge, the WMH volume in 80 WM tracts were calculated (Yeh et al., 2018).
• To avoid computing regressions in tracts with little WMH data, regions that had no WMH voxels in more than 80% of the subjects were discarded, leaving 45 WM tracts with some WMHs in at least 20% of the population.

Deformation Based Morphometry (DBM)

• All the T1-weighted images were nonlinearly registered to the MNI-ICBM152 template using the symmetric diffeomorphic image registration (SyN) tool from ANTS (Avants et al., 2009, 2008).
• Deformation-based morphology (DBM) maps were calculated by computing the Jacobian determinant of the deformation fields obtained from these nonlinear transformations, as a proxy of the relative local volume difference between the individual and MNI-ICBM152 template.
• Similarly, the CerebrA atlas was used to calculate average regional grey matter volume in 102 cortical and subcortical regions (Manera et al., 2020, 2019).
• The CerebrA grey matter atlas is based on the Mindboogle-101 atlas (Klein and Tourville, 2012), which was nonlinearly registered to the MNI-ICBM152 template and manually corrected to remove any remaining partial volume effects.

Cognitive Performance

• The Alzheimer's Disease (AD) Assessment Scale-Cognitive Subscale (ADAS13) scores (Mohs and Cohen, 1987) were used to assess cognitive performance for the ADNI subjects and the Montreal Cognitive Assessment (MoCA) scores (Nasreddine et al., 2005) were used as the cognitive scores of interest for the NIFD and PPMI subjects (no single cognitive score was consistently available for all datasets).
• In each study, the cognitive performance of the disease cohort was compared against the control group from the same study.

Statistical Analysis

• The relationship between regional GM DBM values and regional WMH burden was assessed using the following model for each possible combination of the 102 GM regions and the WM tracts: Regional GM volume ~ 1 + Regional WMH load +.
• The variable of interest in eq. 1 and eq. 2 was Cohort, reflecting the differences between the patients and the appropriate age-matched controls.
• Similarly, the variables of interest in eq. 4 was Regional WMH load:Cohort, reflecting the additional contribution of WMHs to cognitive performance in each cohort.
• All statistical analysis was performed in MATLAB (version R2015b).

Results

• Table 1 provides a summary of the descriptive characteristics for the participants included in this study.
• In the MCI cohort, the results show significant increase over controls in WMH burden predominantly in the fornix, anterior commissure, corpus callosum, bilateral cortico-striatal tract and inferior fronto-occipital vertical occipital fasciculi.
• Colder colors indicate significant shrinkage of the area compared with the ICBM-MNI152-2009c template, i.e. presence of regional atrophy.
• Table S2 in the Supplementary materials shows tstatistics for the top 20 GM regions with greater atrophy for MCI, AD, FTD, PD cohorts compared to their corresponding study age-matched controls.
• The FTD cohort presented with extensive levels of atrophy, more remarkable in the cingulate, deep nuclei (thalamus and putamen) and cortical areas in the frontal and temporal lobes bilaterally .

Discussion

• The authors combined data from three publicly available large databases to investigate the prevalence of regional WMHs in WM tracts and regional GM atrophy, as well as their interplay and impact on cognitive function in three most common neurodegenerative diseases; namely AD, FTD, and PD.
• Other studies investigating later stage PD patients do however report higher incidence of WMHs (Mak et al., 2015; Piccini et al., 1995), substantiating the possibility that the increase might occur at later stages of the disease.
• This might indicate a specific synergistic contribution of the cerebrovascular pathology to the disease-specific patterns of atrophy, as opposed to a nonspecific additional pattern of atrophy over all brain regions.
• One major limitation of the present study was the inconsistencies between the three datasets used.
• WMH burden on WM tracts also correlates with regional grey matter atrophy in pathologically relevant areas (i.e. the frontal lobe for FTD, and diffuse but mainly parietal and temporal lobes for AD).

Figures

Figure 2. Regional grey matter differences between MCI, AD, FTD, PD cohorts and their age-matched controls. Each row includes different axial slices covering the brain, showing the t-statistics for regions that were significantly different between controls and patients after FDR correction. MCI= mild cognitive impairment. AD= Alzheimer’s dementia. FTD= fronto-temporal dementia. PD= Parkinson’s disease. Images presented in neurological format.

Figure 4. Significant interactions between cohort and WMH loads per tract affecting cognitive scores in PD, FTD, MCI, and AD patients (FDR corrected p-value<0.05). Colors indicate the t-statistic values from the mixed effects models (eq. 4), with colder colors indicating poorer cognitive performance relating to higher WMH burden. WMH=White Matter Hyperintensities. PD= Parkinson’s Disease. FTD= Fronto-temporal Dementia. MCI= Mild Cognitive Impairment. AD= Alzheimer’s Dementia. All images in neurological format.

Table 1. Descriptive statistics for the participants enrolled in this study at baseline. Data are number of participants in each category (N), and mean ± standard deviation of the variables.

Figure 3. Significant interactions between WMH loads per tract and cohort, impacting regional GM volumes in MCI, and AD, FTD, and PD patients (Eq. 3). The connections show the significant regions for each tract (FDR corrected pvalue<0.05). Colors indicate t-statistics, with colder colors indicating higher levels of atrophy (lower DBM values) relating to higher WMH burden. WMH=White Matter Hyperintensities. DBM= Deformation Based Morphometry. PD= Parkinson’s Disease. FTD= Fronto-temporal Dementia. MCI= Mild Cognitive Impairment. AD= Alzheimer’s Dementia.

Figure 1. Regional WMH differences between MCI, AD, FTD, PD cohorts and their corresponding study age-matched controls. Each row includes different axial slices covering the brain, showing the t-statistics for regions that were significantly different between controls and patients after FDR correction. WMH= white matter hyperintensities. MCI= mild cognitive impairment. AD= Alzheimer’s dementia. FTD= fronto-temporal dementia. PD= Parkinson’s disease. Images presented in neurological format, i.e. left is on left.

Full text

White Matter Hyperintensities, Grey Matter Atrophy, and Cognitive Decline in
Neurodegenerative Diseases
Mahsa Dadar
1,2
(PhD) mahsa.dadar@mail.mcgill.ca
Ana Laura Manera
1,2
(MD) ana.manera@mail.mcgill.ca
D. Louis Collins
1,2
(PhD) louis.collins@mcgill.ca
1. NeuroImaging and Surgical Tools Laboratory, Montreal Neurological Institute, McGill University, Montreal,
Quebec, Canada.
2. McConnell Brain Imaging Centre, Montreal Neurological Institute, McGill University, Montreal, Quebec,
Canada.
Corresponding Author Information:
Mahsa Dadar, Montreal Neurological Institute, 3801 University Street, Room WB320, Montréal, QC, H3A 2B4
Email: mahsa.dadar@mcgill.ca
.CC-BY-ND 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted April 8, 2021. ; https://doi.org/10.1101/2021.04.06.438619doi: bioRxiv preprint

Abstract:
Introduction: White matter hyperintensities (WMHs) as seen on T2w and FLAIR scans represent
small-vessel disease related changes in the brain. WMHs are associated with cognitive decline in
the normal aging population in general and more specifically in patients with neurodegenerative
diseases. In this study, we assessed the different spatial patterns and relationships between WMHs
and grey matter (GM) atrophy in normal aging, individuals with mild cognitive impairment (MCI),
Alzheimer’s dementia (AD), fronto-temporal dementia (FTD), and de novo Parkinson’s disease
(PD).
Methods: Imaging and clinical data were obtained from 3 large multi-center databases: The
Alzheimer's Disease Neuroimaging Initiative (ADNI), the frontotemporal lobar degeneration
neuroimaging initiative (NIFD), and the Parkinson’s Progression Markers Initiative (PPMI).
WMHs and GM atrophy maps were measured in normal controls (N= 571), MCI (N= 577), AD
(N= 222), FTD (N= 144), and PD (N= 363). WMHs were segmented using T1w and T2w/PD or
FLAIR images and mapped onto 45 white matter tracts using the Yeh WM atlas. GM volume was
estimated from the Jacobian determinant of the nonlinear deformation field required to map the
subject’s MRI to a standard template. The CerebrA atlas was used to obtain volume estimates in
84 GM regions. Mixed effects models were used to compare WMH in different WM tracts and
volume of multiple GM structures between patients and controls, assess the relationship between
regional WMHs and GM loss for each disease, and investigate their impact on cognition.
Results: MCI, AD, and FTD patients had significantly higher WMH loads than the matched
controls. There was no significant difference in WMHs between PD and controls. For each cohort,
significant interactions between WMH load and GM atrophy were found for several regions and
tracts, reflecting additional contribution of WMH burden to GM atrophy. While these associations
were more relevant for insular and parieto-occipital regions in MCI and AD cohorts, WMH burden
in FTD subjects had greater impact on frontal and basal ganglia atrophy. Finally, we found
additional contribution of WMH burden to cognitive deficits in AD and FTD subjects compared
with matched controls, whereas their impact on cognitive performance in MCI and PD were not
significantly different from controls.
Conclusions: WMHs occur more extensively in MCI, AD, and FTD patients than age-matched
normal controls. WMH burden on WM tracts also correlates with regional GM atrophy in regions
anatomically and functionally related to those tracts, suggesting a potential involvement of WMHs
in the neurodegenerative process.
Keywords: White matter hyperintensities, small-vessel disease, neurodegenerative disease,
Alzheimer’s disease, fronto-temporal dementia, Parkinson’s disease, mild cognitive impairment
.CC-BY-ND 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted April 8, 2021. ; https://doi.org/10.1101/2021.04.06.438619doi: bioRxiv preprint

Introduction
White matter hyperintensites (WMHs), defined as nonspecific hyperintense regions in the white
matter tissue of the brain on T2-weighted or FLuid-Attenuated Inversion Recovery (FLAIR)
magnetic resonance images (MRIs) are common findings in the aging population in general
(Hachinski et al., 1987). These age-related WMHs are considered to be the most common MRI
signs of cerebral small vessel disease and are generally due to chronic hypoperfusion and
alterations in the blood brain barrier (McAleese et al., 2016). Other pathological correlates of
WMHs include demyelination, axonal and neuronal loss, higher levels of microglial activation, as
well as arteriosclerosis due to hypoxia, inflammation, degeneration, and amyloid angiopathy
(Abraham et al., 2016; Gouw et al., 2010).
WMHs are known to have a higher incidence in neurodegenerative diseases such as Alzheimer’s
disease (AD) (Capizzano et al., 2004; Dadar et al., 2017a; Dubois et al., 2014; Tosto et al., 2014),
dementia with Lewy bodies (DLB) (Barber et al., 1999), Parkinson’s disease (PD) (Mak et al.,
2015; Piccini et al., 1995), fronto-temporal dementia (FTD) (Varma et al., 2002), as well as
individuals with mild cognitive impairment (MCI) (DeCarli et al., 2001; Lopez et al., 2003; Dadar
et al., 2017a). Patients with WMHs generally present with significantly more severe cognitive
deficits and suffer greater future cognitive decline compared with individuals with the same level
of neurodegeneration related pathologies without WMHs (Au et al., 2006; Carmichael et al., 2010;
Prins and Scheltens, 2015; Dadar et al., 2020b, 2019, 2020a, 2018b, 2020b).
Few studies have investigated the relationship between the longitudinal changes in WMHs in
different white matter tracts, neurodegenerative changes, and cognitive decline. In a relatively
small sample, Burton et al. studied the impact of WMHs in late-life dementia in DLB, PD and AD
(Burton et al., 2006). They found significantly greater total load of WMHs in AD, but not PD or
DLB. They did not find a significant association between the rate of change in WMH load and
cognitive performance (Burton et al., 2006). In a community-based cohort of 519 older adults,
Rizvi et al. found that increased WMH load in association and projection tracts were related to
worse memory function (Rizvi et al., 2020). However, they did not investigate the relationships
with measures of grey matter atrophy. In another aging sample of 2367 adults (age range 20-90
years), Habes et al. reported that WMHs in most tracts were related to age-related atrophy patterns,
as measured by Spatial Pattern of Alteration for Recognition of Brain Aging index (Habes et al.,
.CC-BY-ND 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted April 8, 2021. ; https://doi.org/10.1101/2021.04.06.438619doi: bioRxiv preprint

2018). However, they did not investigate regional grey matter atrophy patterns or the relationships
with cognitive performance.
In this study, we used a previously validated automated WMH segmentation technique (Dadar et
al., 2017c, 2017b) to quantify the WMHs in 3 large multi-center cohorts of neurodegenerative
diseases, with a total of 1730 subjects and 5774 timepoints, and investigated the differences
between spatial distribution of regional WMHs in AD, PD, FTD, MCI, and cognitively normal
individuals. In addition, we investigated the relationship between WM tracts containing WMH
lesions and regional grey matter atrophy and cognitive performance.
Methods
Participants
Data used in this study includes subjects from Alzheimer's Disease Neuroimaging Initiative
(ADNI) database, the Parkinson's Progression Markers Initiative (PPMI), and the frontotemporal
lobar degeneration neuroimaging initiative (NIFD) that had either FLAIR or T2-weighted MR
images.
ADNI
The ADNI (adni.loni.usc.edu) was launched in 2003 as a public-private partnership led by
Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI has been to test
whether serial MRI, positron emission tomography, other biological markers, and clinical and
neuropsychological assessment can be combined to measure the progression of MCI and early AD.
ADNI was carried out with the goal of recruiting 800 adults aged from 55 to 90 years and consists
of approximately 200 cognitively normal patients, 400 patients with MCI, and 200 patients with
AD (http://adni.loni.usc.edu/wp-content/uploads/2010/09/ADNI_GeneralProceduresManual.pdf).
ADNIGO is a later study that followed ADNI participants who were in cognitively normal or early
MCI stages (http://adni.loni.usc.edu/wp-
content/uploads/2008/07/ADNI_GO_Procedures_Manual_06102011.pdf). The ADNI2 study
followed patients in the same categories, recruiting 550 new patients (http://adni.loni.usc.edu/wp-
content/uploads/2008/07/adni2-procedures-manual.pdf). The longitudinal MRI data used in this
study included T1w, T2w/proton density–weighted acquisitions from ADNI1 patients and T1w
and FLAIR acquisitions from ADNI2/GO patients. The scanner information and image acquisition
.CC-BY-ND 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted April 8, 2021. ; https://doi.org/10.1101/2021.04.06.438619doi: bioRxiv preprint

parameters have been previously described (Dadar et al., 2017a). The ADNI1, ADNI2 and
ADNIgo studies acquired data from subjects on a yearly basis.
PPMI
The PPMI (http://www.ppmi-info.org) is a longitudinal multi-site clinical study of approximately
600 de novo PD patients and 200 age-matched healthy controls followed over the course of five
years (Marek et al., 2011). The study was approved by the institutional review board of all
participating sites and written informed consent was obtained from all participants before inclusion
in the study.
NIFD
The frontotemporal lobar degeneration neuroimaging initiative (FTLDNI) is founded through the
National Institute of Aging and started in 2010. The primary goals of FTLDNI are to identify
neuroimaging modalities and methods of analysis for tracking frontotemporal lobar degeneration
(FTLD) and to assess the value of imaging versus other biomarkers in diagnostic roles. The
Principal Investigator of FTLDNI is Dr. Howard Rosen, MD at the University of California, San
Francisco. The data is the result of collaborative efforts at three sites in North America. For up-to-
date information on participation and protocol, please
visit: http://memory.ucsf.edu/research/studies/nifd. The FTLDNI contains 120 cognitively normal
controls and 120 patients with FTD followed yearly for three years.
MRI Measurements
WMHs
All T1-weighted, T2-weighted, proton density (PD), and FLAIR MRI scans were preprocessed in
3 steps using our standardized pipeline: denoising (Manjón et al., 2010), intensity non-uniformity
correction (Sled et al., 1998), and intensity normalization into the range 0–100. For each subject,
the T2-weighted, PD, and FLAIR scans were then co-registered to the T1-weighted scan of the
same visit using a 6-parameter rigid registration and a mutual information objective function
(Collins et al., 1994; Dadar et al., 2018a). Using a previously validated fully automated WMH
segmentation method and a library of manual segmentations based on 53 patients from ADNI1
and 46 patients from ADNI2/GO, the WMHs were automatically segmented for all longitudinal
.CC-BY-ND 4.0 International licenseavailable under a
was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
The copyright holder for this preprint (whichthis version posted April 8, 2021. ; https://doi.org/10.1101/2021.04.06.438619doi: bioRxiv preprint