More atrophy of deep gray matter structures in frontotemporal dementia compared to Alzheimer\'s disease

Poster Presentations: P1 (WT) mice (n¼8) were studied at 17-19 months-of-age. Whole brain perfusion images (360 mm isotropic spatial resolution) were acquired with a multi-slice, pseudocontinuous ASL pulse sequence on a 7T animal MRI system. Mice were scanned both in the awake state and under sevoflurane anesthesia (a known vasodilator) to assess group differences in resting CBF and vasodilatory response, respectively. The perfusion images were co-registered to anatomical MR images acquired during the same session, and spatially normalized to reference coordinate space to generate regional perfusion measures using a fully-automated image processing pipeline (NIGHTWING TM, Biospective Inc.). Results: The Tg mice exhibited a lower resting CBF (w15%) compared to the WT mice in most brain regions. Under sevoflurane, this difference increased to w30% as the Tg mice exhibited an attenuated vasodilatory response relative to the WT group. The coefficient of variation was higher in the awake mice. Conclusions: This study demonstrated that APPoverexpressing Tg mice have impaired resting cerebrovascular function, which is exacerbated under a challenge condition. The greater effect size of the challenge state suggests that utilization of an appropriate vasodilatory challenge could be a useful paradigm in human AD studies. P1-221

RELATIONSHIP BETWEEN CEREBRAL BLOOD FLOW AND CORTICAL LAYER I ASTROCYTES IN A TRANSGENIC MOUSE MODEL OF ALZHEIMER’S DISEASE

Barry J. Bedell1, Marilyn Grand’Maison1, Ming-Kai Ho1, Francois Hebert2, Alex P. Zijdenbos2, 1McGill University, Montreal, Quebec, Canada; 2Biospective Inc., Montreal, Quebec, Canada. Contact e-mail: [email protected] Background: Cerebrovascular dysfunction is an increasingly recognized hallmark of Alzheimer’s disease (AD). However, the relationship between impaired cerebrovascular function and AD-associated cellular alterations remains largely unexplored. Given the essential role that astrocytes play in neurovascular coupling, we sought to assess the relationship between CBF and astrocytes using temporally-matched in vivo arterial spin labeling (ASL) perfusion MR images and post-mortem, quantitative immunohistochemistry (qIHC) measures from a transgenic (Tg) mouse model of AD. Methods: Tg mice overexpressing mutant human amyloid precursor protein (APP Swe/Ind) (n¼22) and age-matched wild-type (WT) mice (n¼24) were studied at both 3.5 and 18.5 months-of-age. Whole brain perfusion images (280 mm isotropic spatial resolution) were acquired with a 3D ASL pulse sequence on a 7T animal MRI system. Regional, cerebral cortical CBF measures were obtained using a fully-automated image processing pipeline (NIGHTWING TM, Biospective Inc.). Following scanning, mice were sacrificed and the entire brain was sectioned. Brain sections underwent glial fibrillary acidic protein (GFAP) IHC staining, digitization, 3D reconstruction, and spatial normalization to the ASL MRI data (PERMITS TM, Biospective Inc). Streamlines running from pial to white matter surfaces were computed across the entire cerebral cortex using an automated image processing method. An intensity profile was generated along each streamline for GFAP qIHC volumes, and these profiles were averaged over cortical regions-of-interest. Correlation analysis was performed between the regional CBF and GFAP measures. Results: A strong correlation was observed between CBF and GFAP-positive astrocytes in Cortical Layer I in both young (r¼0.92) and old (r¼0.79) WT mice. This association was reduced in the young (r¼0.38) and old (r¼0.57) Tg mice. No significant correlations were observed in Cortical Layer VI in any of the groups. Conclusions: This study demonstrated a strong association between superficial GFAPpositive astrocytes and CBF. Further, this relationship was shown to be altered in APP-overexpressing Tg mice. P1-222

THALAMIC SHAPE AND COGNITIVE PERFORMANCE IN AMNESTIC MILD COGNITIVE IMPAIRMENT

Changtae Hahn, The Catholic University of Korea, Seoul, South Korea. Contact e-mail: [email protected]

P385

Background: Although previous postmortem studies have shown that thalamus is involved in early stage of Alzheimer’s disease, no structural neuroimaging studies have conducted in amnestic mild cognitive impairment (aMCI) patients. In addition, the relationships between thalamic deformations and various episodic memory impairments were not clear. The aim of this study was to investigate thalamic shape changes and their relationships with various episodic memory impairments in aMCI. Methods: Thalamic volumes and deformations were compared between the aMCI(N¼32) and the controls(N¼32). In addition, we explored the correlation pattern between thalamic deformations and cognitive dysfunctions in aMCI using a comprehensive neuropsychological battery. Results: Patients with a MCI showed left thalamic atrophy and thalamic deformations in the left dorso-medial and antero-medial areas compared with healthy individuals. Significant correlations were observed between CERAD-K Word List Memory scores and the left dorso-medial areas in aMCI. Verbal delayed recall scores(CERAD-K Word List Recall) were also significantly correlated with the left dorso-medial areas in aMCI. Conclusions: This study was the first to explore the relationships between thalamic deformations and various types of cognitive performances in aMCI. These structural changes in the dorso-medial and pulvinar areas might be at the core of underlying neurobiological mechanisms of thalamic dysfunction and their relevance to verbal and visuospatial delayed recall in aMCI.

P1-223

MORE ATROPHY OF DEEP GRAY MATTER STRUCTURES IN BEHAVIORAL VARIANT FRONTOTEMPORAL DEMENTIA COMPARED TO ALZHEIMER’S DISEASE

Christiane M€oller1, Nikki Dieleman2, Wiesje M. Van der Flier3, Adriaan Versteeg4, Yolande Pijnenburg3, Philip Scheltens5, Frederik Barkhof3, Hugo Vrenken3, 1VU Medical Center, Amsterdam, Netherlands; 2University Medical Center Utrecht, Amsterdam, Netherlands; 3VU University Medical Center, Amsterdam, Netherlands; 4 VU Medical Center, Amsterdam, Netherlands; 5VU University Medical Center, Amsterdam, Netherlands. Contact e-mail: [email protected] Background: The involvement of frontostriatal circuits in behavioral variant Frontotemporal Dementia (bvFTD) suggests that deep gray matter (DGM) structures may be affected in this disease. We therefore investigated whether volumes of DGM structures differed between patients with bvFTD, Alzheimer’s Disease (AD) and controls controls as well as their relationship with cognitive functions and neuropsychiatric symptoms. Methods: We included 24 patients with bvFTD (6368years, 42% females, MMSE 2465), and matched them based on age, gender and educational level at a ratio of 1:3 to 72 AD patients (6368years, 42% females, MMSE 2165) and 72 patients with subjective memory complaints (6368years, 42% females, MMSE 2862). Automated segmentation of 3DT1-weighted images at 3T was used to estimate volumes of hippocampus, amygdala, thalamus, caudate nucleus, putamen, globus pallidus and nucleus accumbens. MANOVA with Bonferroni post-hoc tests was used to compare volumes between groups. In a second model, analyses were repeated while correcting for head size. To assess the relationships between volumes, cognitive functions and neuropsychiatric symptoms within each diagnostic group, we used Spearman correlations. Results: Nucleus accumbens volume discriminated all groups, with bvFTD having most severe atrophy (p¼0.001). Caudate nucleus and globus pallidus volumes discriminated bvFTD from AD (p¼0.004) and from controls (p0.016). Hippocampal volumes only discriminated dementia from controls and amygdala volumes differed solely between bvFTD and controls. Volumes of thalamus and putamen did not differ between groups (Figure 1). Correcting volumes for head size revealed similar results. Correlations between language and volumes of thalamus, caudate nucleus, putamen, globus pallidus and between night-time behavior disturbances and hippocampal and nucleus accumbens volume in controls were found. Hippocampus, amygdala and nucleus accumbens volume were related to memory and hippocampal volume to euphoria in AD. In bvFTD, globus pallidus volume was related to executive functioning,

P386

Poster Presentations: P1

nucleus accumbens volume to disinhibition, and caudate nucleus to nighttime behavior disturbances. Conclusions: Caudate nucleus, globus pallidus and nucleus accumbens were more severely affected in bvFTD than in AD and controls. The observed volume differences of these DGM structures supports the idea that next to frontal cortical atrophy, DGM structures, as parts of the frontal circuits, are damaged in bvFTD rather than in AD.

(positive), with a focal increased uptake in the left temporal-occipital lobe. Conclusions: ARIA-E and associated ARIA-H can be observed in ^ risk allele, cognitively normal elderly who do not carry the APOE O‘4 have no prior microhemorrhages, and are not receiving amyloid-modifying treatments. This is an important consideration when assessing MRI images for safety in amyloid-modifying clinical trials. Focal amyloid deposits around the region of ARIA-H suggest cerebral amyloid angiopathy may be responsible for the occurrence of ARIA in this case. P1-225

CLINICAL MEMORY PERFORMANCE COVARIES WITH CORTICAL THINNING OF THE DEFAULT MODE NETWORK IN ALZHEIMER’S DISEASE: THE SUNNYBROOK DEMENTIA STUDY

Sean Michael Nestor, Jiali Zhao, Mario Masellis, Joel Ramirez, Sandra E. Black, Sunnybrook Research Institute, Toronto, Ontario, Canada. Contact e-mail: [email protected]

Figure 1. Total gray matter volumes (cm3) of hippocampus, amygdala and DGM structures. P1-224

SPONTANEOUS ARIA WITH EDEMA AND MICROHEMORRHAGES IN A COGNITIVELY NORMAL SUBJECT FROM ADNI

Mekala Raman1, Heather Wiste2, Matthew Senjem2, Chad Ward2, Clifford Jack2, Kejal Kantarci2, 1Mayo Graduate School, Rochester, Minnesota, United States; 2Mayo Clinic, Rochester, Minnesota, United States. Contact e-mail: [email protected] Background: Amyloid-Related Imaging Abnormalities (ARIA) include sulcal effusions and edema (ARIA-E) on fluid attenuated inversion recovery (FLAIR) and superficial siderosis and microhemorrhages (ARIA-H) on T2* gradient recalled echo (GRE) images. The risk factors for ARIA-E are higher dosage of amyloid-modifying immunotherapy, carrying the APOE ^ allele, increased clearance of amyloid- b protein, and presence of O‘4 ARIA-H. Risk factors for ARIA-H are cerebral amyloid angiopathy, cardiovascular risk factors, experiencing a cerebrovascular event, and presence of ARIA-E. Observation of spontaneous ARIA in individuals who are not receiving amyloid-modifying treatment is rare. Our objective is to identify the risk of spontaneous ARIA subjects in the Alzheimer’s Disease Neuroimaging Initiative 2 (ADNI2) cohort. Methods: The ADNI2/GO studies have a total of 1006 subjects with a range of diagnoses: cognitively normal, early mild cognitive impairment (MCI), late MCI, Alzheimer’s disease, and significant memory concerns. The subjects have been followed since the start of ADNI2/GO in 2004, with a total of 3385 scans acquired and assessed until a spontaneous ARIA case was identified. Results: The case was an 81-year-old, cognitively normal male who showed MRI evidence of spontaneous ARIA-E on FLAIR images and ARIA-H on T2* GRE images. As a volunteer in the ADNI2 study, he had four serial MRIs for 2 years and had no ARIA until the most recent visit. The findings were identified in the posterior left temporal lobe on MRI during the year 2 visit. The patient was not receiving amyloid-modifying therapy and the APOE genotype was ^ O‘3. ^ O‘3/ The global amyloid SUVR on PET at time of MRI was 1.85

Background: Recent structural and functional magnetic resonance imaging (MRI) studies suggest that neurodegeneration is spatially associated with intrinsic brain functional network architecture, particularly the default mode network (DMN). The multivariate method Partial Least Squares (PLS) is ideal for studying the network degeneration hypothesis in AD, as this technique can map spatial covariance between vertex-wise cortical thickness (CT) and clinical cognitive measures in a single analytical step. Using PLS, we assessed whether a spatial pattern of covariance between CT and memory performance co-localized to DMN architecture in AD versus normal elderly controls (NC). Methods: Cross-sectional data were acquired from the Sunnybrook Dementia Study (SDS): mild AD (n¼144) and aged matched NC (n¼95). SDS subjects were acquired from the cognitive neurology memory clinics at Sunnybrook Health Sciences Centre, Toronto, Canada. All subjects had 1.5 Tesla T1-SPGR MRIs (Matrix¼256x192; TE/TR¼35ms/5ms; flip-angle¼35 , in-plane resolution¼0.85930.859x1.2-1.4mm). An in-house modified version of the Freesurfer Cortical Thickness algorithm (http://ftp.nmr.mgh.harvard.edu/) was used to minimize false positive grey matter signal from white matter hyperintensities. A composite memory score was computed from z-scores for the California Verbal Learning Test total correct at acquisition, Wechsler Memory Scale Revised Visual Reproduction immediate recall and the Dementia Rating Scale memory score. PLS software (McIntosh lab at Rotman Research Institute, Canada) was adapted for surface-based parametric analysis. PLS with bootstrapping (x1000) and permutation testing (x1000) was applied to detect significant group-wise patterns of covariance between CT and memory scores. The Analysis was corrected for sex and age. Results: PLS revealed a significant age-independent pattern of covariance between memory performance and cortical thinning in AD but not NC that co-localized to the posterior cingulate, superior and lateral parietal, medial prefrontal and medial temporal lobe bilaterally (p