Fulvia Palesi1,2, Andrea De Rinaldis2,3, Letizia Casiraghi2,4, Gloria Castellazzi2,3, Paolo Vitali5, Nicoletta Anzalone6, Federica Denaro7, Elena Sinforiani8, Giuseppe Micieli7, Egidio D'Angelo2,4, and Claudia Angela Michela Gandini Wheeler-Kingshott2,9
1Department of Physics, University of Pavia, Pavia, Italy, 2Brain Connectivity Center, C. Mondino National Neurological Institute, Pavia, Italy, 3Department of Electrical, Computer and Biomedical Engineering, University of Pavia, Pavia, Italy, 4Department of Brain and Behavioral Sciences, University of Pavia, Pavia, Italy, 5Brain MRI 3T Mondino Research Center, C. Mondino National Neurological Institute, Pavia, Italy, 6Scientific Institute H. S. Raffaele, Milan, Italy, 7Department of Emergency Neurology, C. Mondino National Neurological Institute, Pavia, Italy, 8Alzheimer's Disease Assessment Unit, Laboratory of Neuropsychology, C. Mondino National Neurological Institute, Pavia, Italy, 9NMR Research Unit, Queen Square MS Centre, Department of Neuroinflammation, UCL Institute of Neurology, University College London, London, United Kingdom
Synopsis
Dementia is the most common disorder in elderly people and comprises Alzheimer’s disease (AD) and vascular disease (VaD). In this work graph theoretical approach was applied to a cohort of AD, VaD and healthy controls (HC) aimed at investigating the presence of a disease-specific pattern of alterations. Brain structural networks were built using the Cohen functional atlas (nodes) and advanced probabilistic tractography (edges). Our main finding was that VaD patients showed severe impairment in the large-scale brain networks while AD patients mainly showed inefficiency of short-range connections emphasizing the fact that alterations are restricted to specific brain regions.Purpose
Dementia is the most common disorder in elderly people and comprises several different subtypes.
1 The extremes of its spectrum are represented by pure Alzheimer’s disease (AD) and vascular disease (VaD). Therefore, distinguishing between AD and VaD is critical because their neuropsychological profiles are similar, while vascular impairment is not correlated with VaD. Recent studies have reported alterations in brain networks of AD and VaD patients suggesting that their impairment follows disease-specific patterns.
2 In this work we combined structural and functional information. In particular, graph theoretical approach was applied to a cohort of AD, VaD and healthy controls (HC) aimed at assessing changes in network integration and segregation with the main purpose of investigating the presence of disease-specific patterns of alterations.
Methods
Subjects: A cohort of 91 subjects including 30 AD (mean age 73±8 yrs, 13 females), 26 VaD (mean age 77±8 yrs, 21 females) and 35 HC (mean age 69±10 yrs, 17 females) were included in the study.
MRI acquisition: All data were acquired using a 3T Skyra scanner (Siemens, Erlangen, Germany) with a 32-channel head-coil. DTI data were acquired using a twice-refocused SE-EPI sequence (TR/TE=10000/97ms, 70 axial slices, FOV=240 mm, 2 mm isotropic voxel, 64 non collinear diffusion directions, b=1200s/mm2 and 10 volumes with b=0 s/mm2). 3DT1w images were collected with a MPRAGE sequence (TR/TE/TI=2300/2.95/900ms, 176 sagittal slices, FOV=270 mm, 1.2x1.1×1.1 mm3 voxel, flip angle=9°).
DTI analysis and tractography: Eddy current correction, brain extraction of the mean b0 image and creation of fractional anisotropy (FA) and mean diffusivity (MD) maps were performed using FSL.3 For each subject, the 3DT1 images were aligned to the corresponding b0 images using a full-affine registration to then obtain tissue-class segmentations in diffusion space. Whole-brain tractography was performed with MRtrix34 by combining constrained spherical deconvolution and probabilistic streamline (iFOD2).5 Default parameters were used: step size=1.25 mm, FA termination=0.1, max angle=45°, 10 million streamlines selected. The boundary between white and grey matter was used for randomly seeding the streamlines. To obtain a valid marker of axonal fibre count, the whole-brain tractography was filtered to 5 million streamlines applying the Spherical-deconvolution Informed Filtering of Tractograms (SIFT6) algorithm.
Structural network analysis: Deep grey matter nuclei and cerebellar hemispheres were added to the Cohen functional atlas7 achieving a brain atlas of 348 parcellations (MNI space), transformed to within subject-space by inverting the non-affine registration from DTI space to MNI space. Three connectivity matrices were created weighting bythe number of streamlines between pairs of nodes, mean FA and mean MD of the tracts. Mean nodal degree, global efficiency, clustering coefficient, and mean strength were calculated. These measurements were compared among groups using a multivariate regression model with age and gender as covariates.
Results
All differences between groups are summarized in
Table 1. VaD showed decrease nodal degree, global efficiency and strength with respect to HC and to AD when weighting for number of streamlines and FA. VaD showed also decrease strength with respect to HC and to AD when weighting for MD. AD showed altered clustering coefficient with respect to HC when weighting for FA and MD. Clustering coefficient weighted by FA was significantly different amongst all groups (HC>AD>VaD). Brain network topology is shown in
Figure 1 for each group of subjects to highlight the progressive disruption of structural connection from HC to AD and to VaD.
Discussion
Our main finding was that VaD patients showed severe impairment in the large-scale brain networks while AD patients mainly showed alterations in local proprieties. In particular, VaD revealed decreases in integration and in segregation, meaning that their networks were less efficient both in long-range and in short-range connections with respect to HC and AD. On the other hand, AD revealed alterations in the clustering coefficient either when using FA or MD as weights. This finding may reflect inefficiency in short-range connections in AD, emphasizing that alterations are restricted to specific brain regions. Furthermore, decreased nodal degree in VaD highlighted that the disruption of axonal fibres was not negligible in these patients, probably due to their severe vascular burden. These observations could be further investigated by restricting the current analysis to simpler brain networks, e.g. each separate resting state network as identified by Cohen
7. Moreover, further studies are warranted to investigate the relationship between structural impairment and cognitive deficit.
Acknowledgements
We thank C. Mondino National Neurological Institute of Pavia (5 per mille 2011) and University of Pavia for funding. References
1. Rizzi L, Rosset I, Roriz-Cruz M. Global epidemiology of dementia: Alzheimer's and vascular types. Biomed Res Int. 2014;2014:908-915.
2. Kim HJ, Im K, Kwon H, et al. Effects of amyloid and small vessel disease on white matter network disruption. J Alzheimers Dis. 2015;44(3):963-75.
3. FMRIB Software Library, http://www.fmrib.ox.ac.uk/fsl/.
4. Brain Research Institute, Melbourne, Australia, https://github.com/Mrtrix3/mrtrix3.
5. Tournier JD, Calamante F, Connelly A. Proc Int Soc Magn Reson Med 2010; p.3163.
6. Smith RE, Tournier JD, Calamante F, et al. SIFT: Spherical-deconvolution informed filtering of tractograms. Neuroimage. 2013;67:298-312.
7. Cohen AL, Fair DA, Dosenbach NUF, et al. Defining functional areas in individual human brains using resting functional connectivity MRI. Neuroimage. 2008;41:45-57.