S5‐02‐02: Networks of regional covariance in MRI gray matter: Reproducible multivariate patterns in Alzheimer's disease, mild cognitive impairment and healthy aging

not available. S5-02-02 NETWORKS OF REGIONAL COVARIANCE IN MRI GRAY MATTER: REPRODUCIBLE MULTIVARIATE PATTERNS IN ALZHEIMER’S DISEASE, MILD COGNITIVE IMPAIRMENT AND HEALTHY AGING Gene E. Alexander, University of Arizona, Tucson, AZ, USA. Contact e-mail: gene.alexander@arizona.edu Background: Neuroimaging studies with magnetic resonance imaging (MRI) have typically relied on univariate analysis methods to investigate regional brain changes associated with healthy aging, Alzheimer’s disease (AD), and the increased risk for AD with mild cognitive impairment (MCI). Multivariate network analyses can test for neuroimaging patterns of spatial covariance, identifying brain changes as network interactions in aging and disease (Alexander and Moeller, 1994). Objective: To demonstrate multivariate network patterns of MRI gray matter that differentially Symposium S5-02: Neuroimaging S168 reflect the regionally distributed effects of healthy aging from MCI and AD. Methods: Applications of multivariate network analyses using the Scaled Subprofile Model (SSM; Moeller et al., 1987) with MRI voxelbased morphometry (VBM) will be presented in studies of healthy aging, amnestic MCI, and AD. As a modified form of principal component analysis, the SSM produces regional patterns and corresponding subject scores that reflect an individual’s pattern expression. Bootstrap re-sampling is performed to determine robust regional contributions to the patterns. Results: SSM analyses with MRI VBM have identified regionally distributed patterns of brain changes associated with healthy aging that show reproducible regional features across samples, with regions in the prefrontal cortex appearing most consistently reduced. Application of this MRI multivariate network analysis method in young and old non-human primates further supports the preferential reduction of prefrontal cortex, in a model of healthy aging in which the full complement of AD pathology does not occur. The identified age-related SSM VBM patterns differ from the SSM pattern reflecting the continuum of clinical severity from normal to amnestic MCI to AD, in which medial and lateral temporal reductions are prominent. In addition, greater expression of this AD-related pattern was associated with the subsequent conversion to dementia in individuals with amnestic MCI. Conclusions: Together, these findings illustrate the potential for using multivariate analyses, like SSM, as a complement to univariate analysis methods to characterize regionally distributed brain changes associated with healthy and pathological aging. The use of this analytic approach may enhance early detection and has the potential to aid in the evaluation of treatments in human and non-human animal models of aging and AD. S5-02-03 SCIENTIFIC AND CLINICAL IMPLICATIONS FOR VIEWING ALZHEIMER’S DISEASE AS A BRAIN NETWORK DISORDER Barry Horwitz, National Institutes of Health, Bethesda, MD, USA. Contact e-mail: horwitzb@mail.nih.gov Several important conceptual developments in recent years in research on brain disorders, including Alzheimer’s disease, are: (1) the emergence of the network paradigm, due in large part to neuroimaging; instead of thinking of the brain as a set of modules (i.e., individual brain regions) that perform specific cognitive functions, the network paradigm argues that cognitive functions are performed by dynamic interactions among different brain areas i.e., by dynamically formed functional networks of brain regions; and(2) the emergence of neural modeling required by the network paradigm, since to deal with the complexity of the dynamic interactions among multiple brain regions, one must employ computational methods. In this talk, I will illustrate several efforts to implement these approaches in the context of functional neuroimaging. In particular, I will show how a network perspective can be used, potentially, for early detection of neurodegenerative disorders. However, I will also discuss the difficulties in interpretating the results of network analyses. S5-02-04 DISRUPTION OF FUNCTIONAL CONNECTIVITY IN CLINICALLY NORMAL OLDER ADULTS HARBORING AMYLOID BURDEN Trey Hedden, Massachusetts General Hospital, Charlestown, MA, USA. Contact e-mail: hedden@nmr.mgh.harvard.edu Background: Recently developed imaging technologies allow the in vivo detection of fibrillar amyloid burden, which appears in a substantial proportion of clinically normal individuals. The presence of amyloid burden in these individuals may have subtle neurological effects despite the absence of clinical symptoms of dementia. We predicted that amyloid burden would be associated with disruption of functional connectivity within a large-scale brain network. Methods: The relationship between fibrillar amyloid burden, measured via positron emission tomography imaging using C-11 Pittsburgh Compound-B (PIB), and functional correlations within a large-scale brain network, measured via functional magnetic resonance imaging (MRI) during rest, was investigated in a cohort of 38 healthy, clinically normal, community dwelling older adults (aged from 60 to 88, M 1⁄4 73.1). Participants were screened using the Clinical Dementia Rating (CDR 1⁄4 0) scale and the Mini-Mental State Examination (minimum score of 27). Fibrillar amyloid burden was defined by specific binding (indexed to a cerebellar reference region) of PIB in gray matter within a large region of interest that included frontal, lateral, and retrosplenial cortices (FLR). Individuals with distribution volume ratios (DVR) within the FLR region of 1.15 or greater were classified as PIB positive, as opposed to PIB negative. Analyses treating DVR within the FLR region as a continuous measure were also conducted and confirmed the findings from the grouped analyses. Functional correlation MRI (fcMRI) was measured among regions of interest within the default network, including a priori defined regions in posterior cingulate, medial prefrontal, and lateral parietal cortices. Functional correlations were computed among these regions for scans during which participants passively viewed a fixation point (rest). Results: PIB negative individuals exhibited significantly higher estimates of fcMRI than PIB positive individuals during rest; in addition, DVR within the FLR region was significantly negatively correlated with fcMRI. The pattern of these results remained unchanged when controlling for chronological age. Exploratory analyses initiated by seeding the posterior cingulate cortex confirmed significant disruption in the default network including functional disconnection of the hippocampal formation. Conclusions: These data suggest that amyloid deposition is related to disruption of the spontaneous coherence of default network activity in clinically normal older adults. S5-02-05 LOSS OF STRUCTURAL, FUNCTIONAL AND EFFECTIVE CONNECTIVITY OF THE VISUAL SYSTEM IN HEALTHY AGING, MILD COGNITIVE IMPAIRMENT AND ALZHEIMER’S DISEASE Arun L. W. Bokde, Trinity College Dublin, Dublin, Ireland. Contact e-mail: Arun.Bokde@tcd.ie Background: In normal aging there are changes in the structural and functional integrity of the brain in perception and higher cognitive functions. Alzheimer’s disease (AD) is in part the result from the alterations in the functional integrity of brain networks that support normal cognition. The alterations was examined in neural networks that support perception and the default mode network. The objective is to detail the changes in the networks supporting visual processing, and the default mode network in normal aging, Mild Cognitive Impaired (MCI) subjects and mild AD patients. We expected the network alterations due to normal aging and AD to differ. Methods: Functional magnetic resonance imaging (fMRI) was used to measure brain activation during performance of two visual tasks (face matching and location matching) and during rest. Standard pre-processing steps were performed on the fMRI data. Brain activation changes in the visual tasks were computed using multiple regression techniques (FSL software package), and the activation of the default mode network was computed using independent components (ICA) analysis. The integrity of a network was calculated using functional connectivity measures and ICA among regions of a network. Results: Along the ventral visual pathway we did not find alterations in brain activation between HC and amdMCI but here were strong functional connectivity changes of the fusiform gyrus. There were decreased functional connectivity in MCI compared to HC in the visual cortex, parietal lobes and dorsal lateral prefrontal cortex. The activated network was also more strongly anti-correlated to the DMN compared to the MCI group. The alterations in functional connectivity between young and older HC during the visual perceptual tasks were between the fusiform gyrus and frontal lobes. In addition, examination of the changes in the DMN in MCI and AD patients compared to HC were located in parietal lobes and