MYELOID MICROVESICLES IN CSF ARE ASSOCIATED WITH MYELIN DAMAGE AND NEURONAL LOSS IN MILD COGNITIVE IMPAIRMENT AND ALZHEIMER DISEASE
Federica Agosta, MD, PhD1, Dacia Dalla Libera, MD1, Edoardo Gioele Spinelli, MD1, Annamaria Finardi, BSci1, Elisa Canu, PhD1, Alessandra Bergami, MLT1, Luisella Bocchio Chiavetto, PhD3, Manuela Baronio, MD4, Giancarlo Comi, MD1,2, Gianvito Martino, MD1, Michela Matteoli, PhD5,6, Giuseppe Magnani, MD1, Claudia Verderio, PhD5, Roberto Furlan, MD, PhD1. 1
INSPE, Division of Neuroscience, Scientific Institute San Raffaele, and 2Vita-Salute San Raffaele
University, via Olgettina 60, 20132 Milano Italy; 3Neuropsychopharmacology Unit, IRCCS Centro S. Giovanni di Dio, Fatebenefratelli, Via Pilastroni 4, 25125, Brescia, Italy; 4Fondazione Poliambulanza, Via Bissolati 56, 25123, Brescia, Italy. 5CNR Institute of Neuroscience and Department of Medical Pharmacology, via Vanvitelli 32, 20129 Milano, Italy, Department of Biotechnology and Translational Medicine; Università di Milano, Via Vanvitelli 32, 20129 Milano, Italy; 6Istituto Clinico Humanitas IRCCS, Milano, Italy.
Correspondence should be addressed to: Roberto Furlan, INSPE, Division of Neuroscience, San Raffaele Scientific Institute, via Olgettina 60, 20132 Milano Italy, E-mail: [email protected].
Running head: Microglia activation in AD and MCI
Number of characters in the title: 131 Number of characters in the running head: 34 Number of words in the abstract: 250 Number of words in the body of the manuscript: 3783 Number of color figures: 4 Number of supplementary figures: 0
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an ‘Accepted Article’, doi: 10.1002/ana.24235
Annals of Neurology
Page 2 of 35
2 Abstract Objectives: We have described cerebrospinal fluid (CSF) myeloid microvesicles (MVs) as marker of microglia activation during neuroinflammation in Alzheimer’s disease (AD), and characterized their ability to produce toxic amyloid β1-42 (Aβ1-42) oligomers from aggregated or soluble substrate. The aim of this study is to investigate the association of CSF myeloid MVs with neuroimaging, clinical, and paraclinical data in AD and mild cognitive impairment (MCI). Methods: We collected CSF from 106 AD patients, 51 MCI patients and 29 neurologically healthy controls. We examined CSF myeloid MVs content and AD markers. A subgroup of 34 AD and 21 MCI patients underwent structural and diffusion tensor (DT) MRI. Results: Higher levels of myeloid MVs were found in the CSF of AD patients and MCI patients converting within 3 years relative to controls, but also, at a lower level, in MCI patients not converting to AD. CSF myeloid MVs were associated with Tau but not with Aβ1-42 CSF levels. CSF MVs levels correlated with white matter (WM) tract damage in MCI, and with hippocampal atrophy in AD. Interpretation: Microglial MVs are neurotoxic and myelinotoxic in the presence of Aβ1-42. CSF myeloid MVs, mirroring microglia activation and MVs release, are associated to WM damage in MCI and hippocampal atrophy in AD. This suggests that hippocampal microglia activation, in the presence of Aβ1-42 in excess, produces neurotoxic and oligodendrotoxic oligomers that, through WM tract damage, spread disease to neighboring and connected areas, causing local microglia activation and propagation of disease through the same sequence of events.
John Wiley & Sons
Page 3 of 35
Annals of Neurology
3 Introduction Inflammation has been hypothesized among the pathogenic mechanisms contributing to Alzheimer’s disease (AD) since the early ‘80ies.1,2 However, the contribution of inflammation, as detrimental, protective, or both, in neurodegenerative diseases, such as AD, is still elusive, leaving the possibility that modulation of inflammation might represent a therapeutic option, despite several failures of anti-inflammatory approaches have been already reported.3 Microglia represent the first line of defense in the central nervous system (CNS), and the first cell type to be activated in case of danger.4 Activated microglia are hypothesized to contribute to AD pathogenesis by promoting neurotoxicity of amyloid β (Αβ) plaques.5 While the absence of neuroinflammation is typical of patients with high plaque burden without dementia,6 activation of microglia correlates to disease development and progression.7-9 On the other hand, microglia have also been described as protective or as changing role, from protective to detrimental, during disease progression.10,11 While secretion of neurotrophic factors and phagocytosis of Αβ mediate beneficial effect of microglia in AD, the molecular mechanisms by which microglia exert neurotoxicity remain largely unknown. We have recently described extracellular membrane microvesicles (MVs), budding from the cell surface, called shed MVs or ectosomes,12 as a novel mechanism of cell-to-cell communication in the brain, through which reactive microglia propagate inflammatory signals.13-15 MVs, which are shed by microglia upon ATP activation14 through the P2X7 receptor localized on lipid rafts,16,17 influence neuronal synaptic activity, enhancing excitatory neurotransmission.18 MVs, produced by microglia exposed in vitro to Αβ1−42, store neurotoxic Αβ forms and are able to convert extracellular Aβ1−42 deposits to neurotoxic soluble forms.5 In vivo production of microglial MVs is high in AD patients and in subjects with mild cognitive impairment (MCI), and AD MVs are toxic for cultured neurons.5 Since shed MVs selectively accumulate various cellular components, including soluble and integral proteins, lipids and nucleic acids, thus reflecting the activation state of donor microglia,19 they represent novel, and potentially highly informative biomarkers of brain inflammation and neurodegeneration in humans.5,15,20 John Wiley & Sons
Annals of Neurology
Page 4 of 35
4 Structural and diffusion tensor (DT) magnetic resonance imaging (MRI) studies have shown that besides the known loss of neurons in the gray matter (GM), subtle brain diffusivity changes occur along white matter (WM) tracts from the prodromal stages of AD, mainly in posterior brain regions, and spread over the course of the disease to involve the frontal lobe.21-23 In AD, but not in MCI21 and cognitively normal elderly who progress to MCI,24 these WM tract alterations are associated to GM atrophy. A growing body of literature suggests that the “active” WM abnormalities observed in the prodromal stages of AD may be secondary to neuroinflammatory factors25 and, thus, constitute an integral part of the degenerative process associated with disease pathophysiology. The combination of MRI findings and CSF biomarkers, such as tau and Aβ1−42 levels, may lead to a better understanding of AD.26 Here we report the first study combining clinical and neuropsychological evaluations, traditional CSF biomarkers, sophisticated MRI assessment, and evaluation of microglial activation through detection of MVs, in a large cohort of AD and MCI patients. We hypothesize that hippocampal microglia produce neurotoxic and oligodendrotoxic oligomers that, through damage of WM tracts, spread disease to neighboring and connected areas.
Methods Subjects Patients with probable AD27 (n=106) and MCI28 (n=51) were enrolled consecutively at the Scientific Institute and University Vita-Salute San Raffaele, Milan (Table 1). After 36 months, 14 MCI patients converted to AD27 (MCI converters). Subjects were excluded if they had: a family history suggestive of an autosomal dominant disease; systemic anti-inflammatory and/or immunosuppressive drug treatment; medical illnesses or substance abuse that interfere with cognitive functioning; other major systemic, psychiatric or neurological illnesses; and other causes of focal or diffuse brain damage at routine MRI, including lacunae and extensive cerebrovascular disorders. An experienced observer, reviewed the severity of cerebrovascular disease according to
John Wiley & Sons
Page 5 of 35
Annals of Neurology
5 the age-related WM change scale;29 subjects above the 90th percentile of the distribution were excluded. All patients underwent clinical evaluation, neuropsychological assessment, and lumbar puncture. CSF from 29 age- and gender-matched healthy donors without known neurological disease (and negative familiar history of neurological diseases) was collected from subjects undergoing local anesthesia for orthopedic surgery (Table 1). A subgroup of 34 AD and 21 MCI patients, that were representative of the larger sample (Supplementary tables e-1 and e-2), underwent structural and DT MRI scans. Thirty age- and gender-matched healthy controls were recruited among spouses of patients and by word of mouth and served as controls for MRI analysis (Table 1). Approval was received from the local ethical standards committee on human experimentation and written informed consent was obtained from all subjects participating in the study or their caregivers.
Cognitive assessment Neuropsychological assessment was performed by an experienced neuropsychologist unaware of CSF and MRI results, who evaluated: global cognitive functioning with the Mini Mental State Examination (MMSE);30 memory function with Rey Auditory Verbal Learning Test (immediate and delayed recall),31 verbal and spatial span32 and Rey’s figure delayed recall test;33 visuo-spatial abilities with the Rey’s Figure Copy Test;33 reasoning and attention functions with the Raven’s coloured progressive34 and the attentive matrices;35 and language with the phonemic and semantic fluency,36 and token test.37 Scores on neuropsychological tests were age-, sex-, and educationcorrected by using related normative values. Scores obtained by patients at neuropsychological evaluation are shown in Table 2.
CSF collection, protein and FACS analysis CSF samples were collected from lumbar puncture in the morning, between 10 to 12 a.m.,Within 4 hours, one CSF aliquot was directly stained, without further processing, with FITC-conjugate John Wiley & Sons
Annals of Neurology
Page 6 of 35
6 Isolectin B4 from Baindeiraea Simplicifolia, (IB4-FITC, SIGMA), and/or annexin-V -APC in 1% BSA, to stain shed MVs of myeloid origin (MMVs). Labelled MMVs were acquired within a fixed time interval with a flow rate of 12 µL per minute on a Canto II HTS flow cytometer (Becton Dickinson). IB4-positive events (number of events/µl) were evaluated as a parameter of MMVs concentration in the CSF. Data were analyzed using FCS 3 software (Beckton Dickinson, Franklin Lakes, NJ, USA). A second aliquot was centrifuged at 800G for 5 min to remove cells and stored frozen for protein analysis. Aβ1-42, total Tau [t-Tau] and phosphorylated Tau [p-Tau] were determined using the Inno-Bia AlzBio3 kit from Fujirebio (Pomezia, Rome).
Cytokine determination To determine inflammatory cytokines, the centrifuged CSF – supernatant left from MMVs analysis- was immediately stored at -80°C until analyzed using Bio-Plex Multiplex Cytokine Assay (Bio-Rad Laboratories, Hercules, CA), according to manufacturer’s instructions. Concentrations of IL-1b, IL-4, IL6, IL10 and TNFα were calculated according to a standard curve generated for each target and expressed as pg/ml. When the concentrations of the cytokines were below the detection threshold, they were assumed to be 0 pg/ml.
MRI study Within one week from lumbar puncture, brain MRI scans were obtained using a 3.0 T scanner (Intera, Philips Medical Systems, Best, the Netherlands). The following sequences were acquired from 34 AD patients, 21 MCI patients and 30 healthy controls: T2-weighted spin echo (SE); fluidattenuated inversion recovery; 3D T1-weighted fast field echo (repetition time [TR]=25 ms, echo time [TE]=4.6 ms, flip angle=30°, field of view [FOV]=230 mm2, matrix=256x256, slice thickness=1 mm, 220 contiguous axial slices, in-plane resolution=0.89x0.89 mm2); and pulsedgradient SE echo planar with sensitivity encoding and diffusion gradients applied in 35 noncollinear directions (acceleration factor= 2.5; TR= 8773 ms; TE= 58 ms; 55 contiguous, 2.3-mm-thick, axial John Wiley & Sons
Page 7 of 35
Annals of Neurology
7 slices; after SENSE reconstruction, the matrix dimension of each slice was 128 × 128, with in-plane pixel size= 1.87 × 1.87 mm, FOV= 231 × 240 mm2; b facto= 900 s/mm2). MRI analysis was performed by an experienced observer, blinded to subjects’ diagnosis and CSF findings. The following MRI analysis were performed: (i) WM hyperintensity load was measured on T2-weighted and FLAIR scans using the Jim software package (Version 5.0, Xinapse Systems, Northants, UK, http://www.xinapse.com); (ii) hippocampal volume was calculated from the T1-weighted images using FIRST, as implemented in the FMRIB software library (FSL) (http://www.fmrib.ox.ac.uk/fsl/first/index.html); (iii) voxel-based morphometry (VBM) to assess the topographical patterns of GM atrophy was performed using SPM8 and the Diffeomorphic Anatomical Registration Exponentiated Lie Algebra registration method;38 (iv) tract-based spatial statistic (TBSS)39 in FSL was performed to investigate the voxel-wise distribution of mean (MD), radial (radD) and axial diffusivities (axD), and fractional anisotropy (FA) abnormalities of the WM. VBM and TBSS procedures were previously described.40
Statistical analysis Demographic, clinical, cognitive, CSF and volumetric data Group differences in categorial variables were assessed using the Fisher’s Exact test. Continuous variables were compared using the Kruskal-Wallis test followed by post-hoc pairwise Dunn’s test or the Mann-Whitney U-test, as appropriate (SAS Release 9.1, SAS Institute, Cary, NC, USA; p value
Federica Agosta, MD, PhD1, Dacia Dalla Libera, MD1, Edoardo Gioele Spinelli, MD1, Annamaria Finardi, BSci1, Elisa Canu, PhD1, Alessandra Bergami, MLT1, Luisella Bocchio Chiavetto, PhD3, Manuela Baronio, MD4, Giancarlo Comi, MD1,2, Gianvito Martino, MD1, Michela Matteoli, PhD5,6, Giuseppe Magnani, MD1, Claudia Verderio, PhD5, Roberto Furlan, MD, PhD1. 1
INSPE, Division of Neuroscience, Scientific Institute San Raffaele, and 2Vita-Salute San Raffaele
University, via Olgettina 60, 20132 Milano Italy; 3Neuropsychopharmacology Unit, IRCCS Centro S. Giovanni di Dio, Fatebenefratelli, Via Pilastroni 4, 25125, Brescia, Italy; 4Fondazione Poliambulanza, Via Bissolati 56, 25123, Brescia, Italy. 5CNR Institute of Neuroscience and Department of Medical Pharmacology, via Vanvitelli 32, 20129 Milano, Italy, Department of Biotechnology and Translational Medicine; Università di Milano, Via Vanvitelli 32, 20129 Milano, Italy; 6Istituto Clinico Humanitas IRCCS, Milano, Italy.
Correspondence should be addressed to: Roberto Furlan, INSPE, Division of Neuroscience, San Raffaele Scientific Institute, via Olgettina 60, 20132 Milano Italy, E-mail: [email protected].
Running head: Microglia activation in AD and MCI
Number of characters in the title: 131 Number of characters in the running head: 34 Number of words in the abstract: 250 Number of words in the body of the manuscript: 3783 Number of color figures: 4 Number of supplementary figures: 0
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an ‘Accepted Article’, doi: 10.1002/ana.24235
Annals of Neurology
Page 2 of 35
2 Abstract Objectives: We have described cerebrospinal fluid (CSF) myeloid microvesicles (MVs) as marker of microglia activation during neuroinflammation in Alzheimer’s disease (AD), and characterized their ability to produce toxic amyloid β1-42 (Aβ1-42) oligomers from aggregated or soluble substrate. The aim of this study is to investigate the association of CSF myeloid MVs with neuroimaging, clinical, and paraclinical data in AD and mild cognitive impairment (MCI). Methods: We collected CSF from 106 AD patients, 51 MCI patients and 29 neurologically healthy controls. We examined CSF myeloid MVs content and AD markers. A subgroup of 34 AD and 21 MCI patients underwent structural and diffusion tensor (DT) MRI. Results: Higher levels of myeloid MVs were found in the CSF of AD patients and MCI patients converting within 3 years relative to controls, but also, at a lower level, in MCI patients not converting to AD. CSF myeloid MVs were associated with Tau but not with Aβ1-42 CSF levels. CSF MVs levels correlated with white matter (WM) tract damage in MCI, and with hippocampal atrophy in AD. Interpretation: Microglial MVs are neurotoxic and myelinotoxic in the presence of Aβ1-42. CSF myeloid MVs, mirroring microglia activation and MVs release, are associated to WM damage in MCI and hippocampal atrophy in AD. This suggests that hippocampal microglia activation, in the presence of Aβ1-42 in excess, produces neurotoxic and oligodendrotoxic oligomers that, through WM tract damage, spread disease to neighboring and connected areas, causing local microglia activation and propagation of disease through the same sequence of events.
John Wiley & Sons
Page 3 of 35
Annals of Neurology
3 Introduction Inflammation has been hypothesized among the pathogenic mechanisms contributing to Alzheimer’s disease (AD) since the early ‘80ies.1,2 However, the contribution of inflammation, as detrimental, protective, or both, in neurodegenerative diseases, such as AD, is still elusive, leaving the possibility that modulation of inflammation might represent a therapeutic option, despite several failures of anti-inflammatory approaches have been already reported.3 Microglia represent the first line of defense in the central nervous system (CNS), and the first cell type to be activated in case of danger.4 Activated microglia are hypothesized to contribute to AD pathogenesis by promoting neurotoxicity of amyloid β (Αβ) plaques.5 While the absence of neuroinflammation is typical of patients with high plaque burden without dementia,6 activation of microglia correlates to disease development and progression.7-9 On the other hand, microglia have also been described as protective or as changing role, from protective to detrimental, during disease progression.10,11 While secretion of neurotrophic factors and phagocytosis of Αβ mediate beneficial effect of microglia in AD, the molecular mechanisms by which microglia exert neurotoxicity remain largely unknown. We have recently described extracellular membrane microvesicles (MVs), budding from the cell surface, called shed MVs or ectosomes,12 as a novel mechanism of cell-to-cell communication in the brain, through which reactive microglia propagate inflammatory signals.13-15 MVs, which are shed by microglia upon ATP activation14 through the P2X7 receptor localized on lipid rafts,16,17 influence neuronal synaptic activity, enhancing excitatory neurotransmission.18 MVs, produced by microglia exposed in vitro to Αβ1−42, store neurotoxic Αβ forms and are able to convert extracellular Aβ1−42 deposits to neurotoxic soluble forms.5 In vivo production of microglial MVs is high in AD patients and in subjects with mild cognitive impairment (MCI), and AD MVs are toxic for cultured neurons.5 Since shed MVs selectively accumulate various cellular components, including soluble and integral proteins, lipids and nucleic acids, thus reflecting the activation state of donor microglia,19 they represent novel, and potentially highly informative biomarkers of brain inflammation and neurodegeneration in humans.5,15,20 John Wiley & Sons
Annals of Neurology
Page 4 of 35
4 Structural and diffusion tensor (DT) magnetic resonance imaging (MRI) studies have shown that besides the known loss of neurons in the gray matter (GM), subtle brain diffusivity changes occur along white matter (WM) tracts from the prodromal stages of AD, mainly in posterior brain regions, and spread over the course of the disease to involve the frontal lobe.21-23 In AD, but not in MCI21 and cognitively normal elderly who progress to MCI,24 these WM tract alterations are associated to GM atrophy. A growing body of literature suggests that the “active” WM abnormalities observed in the prodromal stages of AD may be secondary to neuroinflammatory factors25 and, thus, constitute an integral part of the degenerative process associated with disease pathophysiology. The combination of MRI findings and CSF biomarkers, such as tau and Aβ1−42 levels, may lead to a better understanding of AD.26 Here we report the first study combining clinical and neuropsychological evaluations, traditional CSF biomarkers, sophisticated MRI assessment, and evaluation of microglial activation through detection of MVs, in a large cohort of AD and MCI patients. We hypothesize that hippocampal microglia produce neurotoxic and oligodendrotoxic oligomers that, through damage of WM tracts, spread disease to neighboring and connected areas.
Methods Subjects Patients with probable AD27 (n=106) and MCI28 (n=51) were enrolled consecutively at the Scientific Institute and University Vita-Salute San Raffaele, Milan (Table 1). After 36 months, 14 MCI patients converted to AD27 (MCI converters). Subjects were excluded if they had: a family history suggestive of an autosomal dominant disease; systemic anti-inflammatory and/or immunosuppressive drug treatment; medical illnesses or substance abuse that interfere with cognitive functioning; other major systemic, psychiatric or neurological illnesses; and other causes of focal or diffuse brain damage at routine MRI, including lacunae and extensive cerebrovascular disorders. An experienced observer, reviewed the severity of cerebrovascular disease according to
John Wiley & Sons
Page 5 of 35
Annals of Neurology
5 the age-related WM change scale;29 subjects above the 90th percentile of the distribution were excluded. All patients underwent clinical evaluation, neuropsychological assessment, and lumbar puncture. CSF from 29 age- and gender-matched healthy donors without known neurological disease (and negative familiar history of neurological diseases) was collected from subjects undergoing local anesthesia for orthopedic surgery (Table 1). A subgroup of 34 AD and 21 MCI patients, that were representative of the larger sample (Supplementary tables e-1 and e-2), underwent structural and DT MRI scans. Thirty age- and gender-matched healthy controls were recruited among spouses of patients and by word of mouth and served as controls for MRI analysis (Table 1). Approval was received from the local ethical standards committee on human experimentation and written informed consent was obtained from all subjects participating in the study or their caregivers.
Cognitive assessment Neuropsychological assessment was performed by an experienced neuropsychologist unaware of CSF and MRI results, who evaluated: global cognitive functioning with the Mini Mental State Examination (MMSE);30 memory function with Rey Auditory Verbal Learning Test (immediate and delayed recall),31 verbal and spatial span32 and Rey’s figure delayed recall test;33 visuo-spatial abilities with the Rey’s Figure Copy Test;33 reasoning and attention functions with the Raven’s coloured progressive34 and the attentive matrices;35 and language with the phonemic and semantic fluency,36 and token test.37 Scores on neuropsychological tests were age-, sex-, and educationcorrected by using related normative values. Scores obtained by patients at neuropsychological evaluation are shown in Table 2.
CSF collection, protein and FACS analysis CSF samples were collected from lumbar puncture in the morning, between 10 to 12 a.m.,Within 4 hours, one CSF aliquot was directly stained, without further processing, with FITC-conjugate John Wiley & Sons
Annals of Neurology
Page 6 of 35
6 Isolectin B4 from Baindeiraea Simplicifolia, (IB4-FITC, SIGMA), and/or annexin-V -APC in 1% BSA, to stain shed MVs of myeloid origin (MMVs). Labelled MMVs were acquired within a fixed time interval with a flow rate of 12 µL per minute on a Canto II HTS flow cytometer (Becton Dickinson). IB4-positive events (number of events/µl) were evaluated as a parameter of MMVs concentration in the CSF. Data were analyzed using FCS 3 software (Beckton Dickinson, Franklin Lakes, NJ, USA). A second aliquot was centrifuged at 800G for 5 min to remove cells and stored frozen for protein analysis. Aβ1-42, total Tau [t-Tau] and phosphorylated Tau [p-Tau] were determined using the Inno-Bia AlzBio3 kit from Fujirebio (Pomezia, Rome).
Cytokine determination To determine inflammatory cytokines, the centrifuged CSF – supernatant left from MMVs analysis- was immediately stored at -80°C until analyzed using Bio-Plex Multiplex Cytokine Assay (Bio-Rad Laboratories, Hercules, CA), according to manufacturer’s instructions. Concentrations of IL-1b, IL-4, IL6, IL10 and TNFα were calculated according to a standard curve generated for each target and expressed as pg/ml. When the concentrations of the cytokines were below the detection threshold, they were assumed to be 0 pg/ml.
MRI study Within one week from lumbar puncture, brain MRI scans were obtained using a 3.0 T scanner (Intera, Philips Medical Systems, Best, the Netherlands). The following sequences were acquired from 34 AD patients, 21 MCI patients and 30 healthy controls: T2-weighted spin echo (SE); fluidattenuated inversion recovery; 3D T1-weighted fast field echo (repetition time [TR]=25 ms, echo time [TE]=4.6 ms, flip angle=30°, field of view [FOV]=230 mm2, matrix=256x256, slice thickness=1 mm, 220 contiguous axial slices, in-plane resolution=0.89x0.89 mm2); and pulsedgradient SE echo planar with sensitivity encoding and diffusion gradients applied in 35 noncollinear directions (acceleration factor= 2.5; TR= 8773 ms; TE= 58 ms; 55 contiguous, 2.3-mm-thick, axial John Wiley & Sons
Page 7 of 35
Annals of Neurology
7 slices; after SENSE reconstruction, the matrix dimension of each slice was 128 × 128, with in-plane pixel size= 1.87 × 1.87 mm, FOV= 231 × 240 mm2; b facto= 900 s/mm2). MRI analysis was performed by an experienced observer, blinded to subjects’ diagnosis and CSF findings. The following MRI analysis were performed: (i) WM hyperintensity load was measured on T2-weighted and FLAIR scans using the Jim software package (Version 5.0, Xinapse Systems, Northants, UK, http://www.xinapse.com); (ii) hippocampal volume was calculated from the T1-weighted images using FIRST, as implemented in the FMRIB software library (FSL) (http://www.fmrib.ox.ac.uk/fsl/first/index.html); (iii) voxel-based morphometry (VBM) to assess the topographical patterns of GM atrophy was performed using SPM8 and the Diffeomorphic Anatomical Registration Exponentiated Lie Algebra registration method;38 (iv) tract-based spatial statistic (TBSS)39 in FSL was performed to investigate the voxel-wise distribution of mean (MD), radial (radD) and axial diffusivities (axD), and fractional anisotropy (FA) abnormalities of the WM. VBM and TBSS procedures were previously described.40
Statistical analysis Demographic, clinical, cognitive, CSF and volumetric data Group differences in categorial variables were assessed using the Fisher’s Exact test. Continuous variables were compared using the Kruskal-Wallis test followed by post-hoc pairwise Dunn’s test or the Mann-Whitney U-test, as appropriate (SAS Release 9.1, SAS Institute, Cary, NC, USA; p value
