Progression rates from mild cognitive impairment to dementia by biomarker and memory thresholds

Progression rates from mild cognitive impairment to dementia by biomarker and memory thresholds

Poster Presentations: P3 and also within ICN-specific cortical regions. Results: We found that even in the absence of any cognitive or structural cha...

74KB Sizes 4 Downloads 33 Views

Poster Presentations: P3

and also within ICN-specific cortical regions. Results: We found that even in the absence of any cognitive or structural changes, global cortical Ab burden was associated with functional connectivity in the DMN as well as cortical networks involved in executive control, motor and perceptual timing, and visual detection but not salience processing, attention, and working memory. We also found that there were significant effects of local network Ab burden (i.e., ICN Ab burden) on the network functional connectivity for all ICNs considered in this study; and the effects were associated more closely to local network Ab burden than global Ab burden. Conclusions: The relationship between local network Ab burden and disrupted intrinsic connectivity in various brain networks, in the absence of cognitive deficit and brain atrophy (and presumably absence of cortical tau tangles and neurodegeneration based on previous pathology reports), suggests that neuronal dysfunction may be due to local toxicity of Ab independent of the presence of tau. Furthermore, the results suggest that local toxicity of Ab may represent an early change in preclinical AD.

P3-166

METABOLIC RISK FACTOR BURDEN AND CORTICAL THICKNESS IN THE MESIAL TEMPORAL LOBE

Claire Murphy1,2,3, Elissa McIntosh2, Erin Green1, Aaron Jacobson2, Lori Haase1, Nobuko Kemmotsu4, 1SDSU/UCSD Joint Doctoral Program, San Diego, CA, USA; 2San Diego State University, San Diego, CA, USA; 3 University of California, San Diego, La Jolla, CA, USA; 4San Diego State University Research Foundation, San Diego, CA, USA. Contact e-mail: [email protected] Background: Metabolic syndrome (MetS) involves a constellation of risk factors for cardiac and vascular disease that are significantly associated with cognitive decline and dementia in late life. The prevalence of MetS is rising and increases with age to more than 45% of adults > 60 years old. This study sought to identify neural correlates of metabolic risk factor burden in non-demented middleaged and older adults, specifically in mesial temporal lobe areas vulnerable to early Alzheimer’s disease. Methods: Participants were 26 middle aged and older adults who ranged in metabolic risk factor burden. The International Diabetes Federation guidelines include the following as contributors to metabolic risk factor burden: insulin resistance, dyslipidemia (elevated triglyceride and low high-density lipoprotein [HDL] cholesterol levels), central obesity, elevated blood pressure, and impaired glucose tolerance or diabetes mellitus. Metabolic risk factor burden was operationally defined by the sum of MetS criteria met by a participant. High-resolution T1-weighted structural MRI scans with prospective motion correction were acquired on a 3T scanner and processed using the Freesurfer image analysis suite. Age and gender were covariates in analyses. Results: Pearson correlations between cortical thickness estimates in mesial temporal lobe areas and metabolic risk factor burden were computed. Cortical thickness was negatively associated with metabolic risk factor burden in left entorhinal cortex, left parahippocampal gyrus, left temporal pole, and right temporal pole. Conclusions: The data revealed significant relationships between metabolic risk factor burden and cortical thickness in areas vulnerable to prodromal Alzheimer’s disease (notably, entorhinal cortex). We hypothesize that changes in mesial temporal lobe thickness associated with metabolic syndrome will contribute to cognitive decline in later life. The findings suggest the importance of further investigation of middle-aged and older adults who carry a heavy metabolic risk burden and thus may be at increased risk

P693

for cognitive decline and dementia. Supported by NIH grant AG004085-26 to CM. The authors have no conflicts of interest to declare. We thank the members of the Lifespan Human Senses Laboratory, and Dr. Thomas Liu and the staff of the UCSD Center for Functional MRI. P3-167

PREDICTING AND CHARACTERIZING DEVELOPMENT OF PRECLINICAL ALZHEIMER’S DISEASE

Matthew R. Brier1, John E. McCarthy2, Tammie L. S. Benzinger2, Yi Su2, Karl A. Friedrichsen2, John C. Morris1, Beau Ances2, Andrei Vlassenko2, 1 Washington University School of Medicine, Saint Louis, MO, USA; 2 Washington University in St. Louis, St. Louis, MO, USA. Contact e-mail: [email protected] Background: Preclinical Alzheimer’s disease (AD) can be operationally defined as an abnormal amyloid-imaging scan. However, determination of abnormality is often made using a single scalar summary metric that exceeds some threshold. The selection of this threshold is arbitrary and some number of individuals who have normal scans will develop preclinical AD in the future. This work aims to identify those persons at high risk of future conversion and to characterize the longitudinal amyloid accumulation in those who develop preclinical AD. Methods: Regional PIB binding was measured within FreeSurfer defined anatomical regions of interest from 131 cognitively normal (clinical dementia rating ¼ 0) and had PIB scans defined as normal using mean cortical SUVR corrected for partial volume effects. At follow up (approx. 3 years later), 16 participants demonstrated positive PIB scans. Penalized linear regression was used to predict conversion status using only the first scan and to identify brain regions predictive of that conversion. Amyloid accumulation was investigated using the longitudinal scan and a canonical correlation approach to describe the topography of PIB accumulation. Results: Penalized regression was able to predict PIB conversion in the future more accurately than previously defined metrics in the independent validation cohort. The evolution of PIB topography longitudinally consisted of a local and distributed process. Some brain regions accumulated amyloid locally at the longitudinal time point. However, other brain regions were predictive of accumulation in distant brain regions at follow up. Conclusions: The proposed approach is well able to identify individuals who are amyloid negative at baseline but who become amyloid positive in the future. This suggests that there is a reliable pathological process that precedes overt detectable disease. Identification of these individuals is critical for the trial of disease modifying therapies. Further, these results support models of amyloid spread in the process of AD progression.

P3-168

PROGRESSION RATES FROM MILD COGNITIVE IMPAIRMENT TO DEMENTIA BY BIOMARKER AND MEMORY THRESHOLDS

Warren W. Barker1, Maria Greig-Custo1, Rosemarie Rodriquez1, David Loewenstein2, Malek Adjouadi3, Mohammed Goryawala3, Qi Zhou3, Ranjan Duara1,2,3, 1Mount Sinai Medical Center, Miami Beach, FL, USA; 2 University of Miami School of Medicine, Miami, FL, USA; 3Florida International University, Miami, FL, USA. Contact e-mail: Ranjan.Duara@ msmc.com Background: Amyloid load, hippocampal atrophy and memory impairment are associated with increased risk for progression from MCI to dementia. However, concrete estimates of the risk conferred by each of these factors, alone and in combination, are

P694

Poster Presentations: P3

not readily available to clinicians, or for counseling patients and their families. Methods: Data for 513 elderly MCI subjects with amyloid PET and brain MRI scans, followed for up to 4 years (mean¼2.460.9yrs), was downloaded from the ADNI database (adni. loni.usc.edu; downloaded 27Jan2015). At baseline, the subjects were classified as amyloid positive (Amy+) or negative (Amy-), hippocampal atrophy positive (HP+) or negative (HP-), or memory impaired (Mem+) or unimpaired (Mem-). Thresholds for Amy+ was an SUVR1.1, for HP+ was hippocampal volume (R+L)/ ICV  0.423%, and Mem+ was an ADAS13 score of  15. Cox regression analyses were used to determine predictors of progression to dementia and Kaplan-Meier estimates of progression rates were computed. Results: 91 MCI subjects progressed to dementia, but neither age nor education predicted progression. In univariate Cox models, amyloid load, normalized hippocampal volume and ADAS13 scores predicted progression. In multivariate Cox regression, the continuous predictors were amyloid load (c2 ¼14.9, p<.0001) and ADAS13 (c2¼56.5, p<.0001), which were additive, but not synergistic. Similar results were obtained using dichotomized predictors. The estimated 3.1 year progression rates to dementia and 95% Confidence Intervals were: Amy+: 54% (4464%); Amy-: 9% (4-19%); HP+: 52% (40-65%); HP-: 23% (1632%); Mem+: 67% (55-78%); Mem-: 12% (7-21%); Amy- and HP-: 4% (1-15%); Amy+ and HP+: 64% (51-77%); HP+ and Mem+: 75% (59-89%); Amy+ and Mem+: 76.% (64-87%); Amy+ and Mem+ and HP: 81% (65-93%). Conclusions: Among ADNI participants diagnosed with MCI, the likelihood of progression to dementia in a three-year period is as low as about 5%, or as high as 75% or greater, based on combinations of biomarkers and memory scores above or below a threshold of impairment. Those with memory impairment and a positive biomarker are at the greatest risk for progression to dementia.

P3-169

REDUCED DEFAULT NETWORK FUNCTIONAL CONNECTIVITY AND VERBAL LEARNING IN COGNITIVELY UNIMPAIRED LATE MIDDLEAGED AND OLDER ADULTS: EXPLORATORY FINDINGS FROM THE ARIZONA APOE COHORT STUDY

Irene R. Beck1, Kewei Chen1, Auttawut Roontiva1, Tonita Emily Wroolie2, Robert Bauer, III,1, Cole Dunbar3, Eric Peshkin4, Dan Bandy1, Nathalie Rasgon2, Richard J. Caselli5, Eric M. Reiman1, 1Banner Alzheimer’s Institute, Phoenix, AZ, USA; 2Stanford University, Stanford, CA, USA; 3Columbia University, New York, NY, USA; 4Duke University, Durham, NC, USA; 5Mayo Clinic Airzona, Scottsdale, AZ, USA. Contact e-mail: [email protected] Background: Epidemiological studies have implicated that metabolic syndrome (defined as the presence of 3 out of 5 factors: abdominal obesity, elevated triglycerides, reduced HDL-C, elevated blood pressure, and elevated fasting glucose) is associated with the risk of age-related cognitive decline and the clinical stages of Alzheimer’s Disease (AD). In this study, we explored the possibility that the metabolic syndrome is associated with altered default mode network (DMN) functional connectivity, memory and learning in cognitively unimpaired late middle-aged and older adults with two, one and no copies of the apolipoprotein E (APOE4) allele, the major genetic risk factor for late-onset AD. Methods: Resting state functional connectivity magnetic resonance images (fcMRIs) and auditory verbal learning test (AVLT) longterm recall, immediate recall, and learning test scores were

analyzed in 98 cognitively unimpaired 6564 year old adults with a reported first degree family history of dementia, including 16 APOE4 homozygotes, 26 heterozygotes, and 56 non-carriers with similar ages, gender distributions, and educational levels. A DMN map was created from each person’s fcMRI using the right angular gyrus as the seed region of interest. Results: Metabolic syndrome was associated with reduced resting state functional connectivity of several DMN regions, including right precuneus, lateral parietal cortex, right and left hippocampal, lateral temporal, frontal and occipital cortex locations, and with increased connectivity in other DMN regions, including temporal, frontal, and occipital locations (P<0.001, uncorrected for multiple regional comparisons). It was also associated with reduced AVLT learning scores (P¼0.03, uncorrected for the 3 AVLT comparisons). Functional connectivity, memory and learning findings were not significantly associated with APOE4 gene dose. Conclusions: While our findings should be considered exploratory, they support the possibility that metabolic syndrome is associated with age-related declines in learning and the risk for AD, and they illustrate how functional connectivity MRI can be used as a preclinical endophenotype of AD.

P3-170

REPRODUCIBILITY OF HIPPOCAMPAL ATROPHY RATE AT 1.5T AND 3T FOR FREESURFER AND MAPS-HBSI USING THE ADNI1 DATA SET

Keith S. Cover1, Ronald A. van Schijndel1, Adriaan Versteeg1, Kelvin K. Leung2, Alberto Redolfi3, Soheil Damangir4, Paolo Bosco3, Jer^ome Revillard5, Baptiste Grenier5, David Manset5, Hugo Vrenken1, Bob W. van Dijk1, Nick C. Fox6, Giovanni Battista Frisoni7,8, Frederik Barkhof1 neuGRID1VU University Medical Center, Amsterdam, Netherlands; 2University College London, London, United Kingdom; 3 IRCCS Fatebenefratelli, Brescia, Italy; 4Karolinska Institutet, Stockholm, Sweden; 5gnubila France, Archamps, France; 6UCL Institute of Neurology, London, United Kingdom; 7IRCCS Istituto Centro San Giovanni di Dio Fatebenefratelli, Brescia, Italy; 8University Hospitals and University of Geneva, Geneva, Switzerland. Contact e-mail: [email protected] Background: Previous results for measuring hippocampal atrophy rates demonstrated MAPS-HBSI has about 50% better reproducibility than FreeSurfer at 1.5T for N¼562 subjects from the ADNI1 data set. The purpose of the current study is to compare their reproducibilities at 3T. Methods: While rarely mentioned in the literature, ADNI1 acquired a second back-to-back (BTB) 3D T1 weighted MPRAGEs at each patient visit. The two BTB MPRAGEs at each patient visit enabled two hippocampal atrophy rate measurements to be calculated for each subject between the baseline and year 1 patient visits. The difference of the two atrophy rates provided a measure of the reproducibility for each subject. ADNI1 acquired BTB MPRAGEs at baseline and year 1 at both 3T and 1.5T for N¼111 subjects. Atrophy rates were calculated using both FreeSurfer 5.3.0 in longitudinal mode and MAPS-HBSI. The atrophy rate is measured as the percentage change in volume over 1 year and the units of reproducibility of each subject is percentage points. Results: The figure shows scatter plots for both FreeSurfer (a) and MAPS-HBSI (b) of the reproducibility of each subject for 3T versus 1.5T for the left hippocampus. The larger scatter of FreeSurfer both vertically (3T) and horizontally (1.5T) demonstrates MAPS-HBSI has better reproducibility at both 3T and 1.5T. Also, FreeSurfer has better reproducibility at 1.5T than at 3T. For the left hippocampus, the reproducibility of MAPS-HBSI is better than FreeSurfer at 3T (p<0.00001) by 87% and at 1.5T (p¼0.004) by 50%. FreeSurfer