P636
Poster Presentations: P3
algorithms that incorporate other clinical data as well as their potential utility in predicting conversion to MCI or dementia. P3-220
WAYFINDING IN ELDERLY AND ALZHEIMER’S DISEASE POPULATIONS: A STUDY IN THE VR-MAZE TEST
Francesca Morganti1, Giuseppe Riva2, 1University of Bergamo, Bergamo, Italy; 2Catholic University of Milan, Milan, Italy. Contact e-mail:
[email protected] Background: Recent neuroimaging studies suggest that hippocampal and parahippocampal systems, toghether with the posterior cingulate region, are associated with the risk factor of Alzheimer’s disease and show significant metabolism changes during Alzheimer’s disease early stages. According to this framework we would like to study whether in Alzheimer population there is a decline in performing the "allocentric to egocentric" translation that requires the contribution of hippocampal and parahippocampal regions. Methods: In this study we evaluated in virtual reality 26 healthy volunteers (Mean age 77.23; sd 5.25) and 26 Alzheimer’s patients (Mean age 80.96; sd 6.3). In order to test participants’ ability to explore a complex environment in an egocentric way by using an allocentric map we used the VR-Maze test, an originally developed virtual reality tool (Morganti et al. 2009) based on 5 differently complex virtual reality mazes. In the VRMaze test, participants were requested to first perform the paper and pencil (PP) version of the mazes, and then to find the right way into the equivalent virtual reality (VR) version of the mazes. Results: A first comparison between Alzheimer and healthy subject was performed on the correct execution of each maze both for the PP and VR version. For PP mazes a between groups analysis showed a significant difference among Alzheimer’s and Healthy subjects only for the more complex maze: maze 5 F(5.26), p¼ .024. For the VR mazes the same analysis revealed an highly significant difference between groups for 4 of the 5 mazes: maze 2 F(4.48), p ¼ .039; maze 3 F (5.95), p ¼ .018; maze 4 F(4.54), p ¼ .038; maze 3 F(4.5), p ¼ .036. Conclusions: Results pointed out how a decreasing ability in performing VR mazes exists in Alzheimer’s patients while this decrease is not so evident in the PP version of the test. These results highlights how there is a reduction in Alzheimer’s patients in the allocentric/egocentric translation demanded from the VR maze test. We can partially conclude that this tool was able to shed light on the differences between a non pathological cognitive decline in elderly and an impairment on this ability linkable with Alzheimer’s disease. P3-221
A PROPORTIONAL ODDS APPROACH FOR ESTIMATING PROBABILITY OF COGNITIVE HEALTH, MCI AND DEMENTIA ASSOCIATED WITH COMPUTERIZED COGNITIVE TESTING SCORE
Glen Doniger1, David Zucker2, Ely Simon1, 1NeuroTrax Corporation, Bellaire, Texas, United States; 2Hebrew University, Jerusalem, Israel. Contact e-mail:
[email protected] Background: With the advent of in-office computerized tools for fine measurement of neurocognitive function, methods are needed for estimating probability of disease from an overall score. Traditional binary cutoffs are necessarily arbitrary and do not accommodate tripartite diagnostic classification or capture the range of probabilities across all possible scores. The current study describes application of the proportional odds model to estimate the probability of cognitive health, MCI, and dementia for each possible value of a summary score from in-office computerized neurocognitive testing. Methods: Computerized neurocognitive testing data were drawn retrospectively from the NeuroTrax database for research participants over age 50 with a diagnosis of cognitively healthy (N¼160), MCI (N¼325) or dementia (N¼75) who completed the MindStreams Global Assessment Battery, a 4560 minute battery for mild impairment assessing seven cognitive domains. Proportional odds (ordinal logistic) regression was used to model probability of each diagnosis for each value of the Global Cognitive Score (GCS), an age- and education-adjusted measure of overall performance scored on an IQ-style scale. Results: Distinct probability functions were generated for cognitively healthy, dementia, and MCI, with highest probability at high
scores for cognitively healthy, low scores for dementia, and intermediate scores for MCI. Probability of cognitive health rose as scores increased, from 0.01 at scores 55-62 to 0.10 at 81, 0.80 at 116, and 0.90 at scores >123. Conversely, probability of dementia fell as scores increased, from 0.78 at a score of 55 to 0.10 at 88 to 0.00 at scores >118. Probability of MCI rose and then fell as scores increased, rising from 0.21 at a score of 55 to a peak probability of 0.72 at 85, back down to 0.21 at 115, and continuing down to 0.01 at scores 142-145. Conclusions: Proportional odds represents an advance over traditional binary cutoffs in that it better reflects the clinical reality of a continuum of probability over the score range and assigns probabilities for multiple diagnoses to a single score. Such an approach may be incorporated into a clinical algorithm to improve the diagnostic process. Follow-up studies are needed to translate the proportional odds methodology into practical clinical tools and evaluate their feasibility. P3-222
DETECTION OF MILD COGNITIVE IMPAIRMENT AND EARLY-STAGE DEMENTIA WITH AN AUDIORECORDED COGNITIVE SCALE
Margaret Sewell1, Xiaodong Luo2, Judith Neugroschl2, Mary Sano3, Icahn School of Medicine at Mount Sinai, New York, New York, United States; 2Mount Sinai School of Medicine, New York, New York, United States; 3Mount Sinai School of Medicine & James J. Peters VAMC, New York, New York, United States. Contact e-mail:
[email protected] 1
Background: Primary care physicians often miss a diagnosis of Mild Cognitive Impairment (MCI) or early dementia and screening measures can be insensitive to very mild impairments. Other cognitive assessments may take too much time or be unacceptable or frustrating to seniors. This study examined the ability of an audio-recorded scale, developed in Australia, to detect MCI or mild Alzheimer’s disease (AD) and compared cognitive domain specific performance on the audio-recorded scale to the in-person battery and common cognitive screens. Methods: Seventy-six subjects from the Mount Sinai Alzheimer’s Disease Research Center were recruited. Subjects were 75 years or older, with a clinical diagnosis of AD or MCI (n¼51) or normal control (n¼25). Participants underwent in-person testing with the neuropsychological testing battery followed within six months by testing with the Audio-recorded Cognitive Screen (ARCS). The ARCS is administered using a compact-disk recorder and headphones, with the subject writing responses in a booklet. Results: The ARCS provided better discrimination between normal and impaired elders than the Mini-Mental Status Exam (MMSE), the clock drawing test or the two combined. The in-person battery and ARCS analogous variables were significantly correlated, most in the .4 to .7 range, including verbal memory, executive function/attention, naming, and verbal fluency. The area under the curve generated from ROC curves indicated high and equivalent diagnostic discrimination for ARCS and the in-person battery (0.972 vs. 0.988; p¼0.23). Conclusions: The ARCS demonstrated better discrimination between normal controls and those with mild cognitive deficits than typical screening measures. Performance on cognitive domains within the ARCS was well correlated with the in-person cognitive test battery. Successful completion of the ARCS was accomplished despite mild difficulty hearing the instructions even in very elderly subjects. The ARCS, which can be administered to unsupervised patients by untrained personnel, represents something of a "hybrid" instrument: longer than typical screening measures and shorter than a standard neuropsychological battery, indicating that it may be a useful measure in primary care settings. P3-223
CLINICAL UTILITY OF THE COGSTATE BRIEF BATTERY IN ALZHEIMER’S DISEASE–RELATED MEMORY IMPAIRMENT
Paul Maruff1, Yen Ying Lim2, David Ames3, Kathryn Ellis4, Robb Pietrzak5, Greg Savage6, Karra Harrington7, Ashley Bush8, Cassandra Szoeke9, Ralph Martins10, Victor Villemagne11, Christopher Rowe12, David Darby4, Colin Masters4, 1CogState Ltd., Melbourne, Australia; 2University of Melbourne, Parkville, Australia; 3 National Ageing Research Institute Inc. (NARI), Parkville, Australia; 4 Mental Health Research Institute, Parkville, Australia; 5Yale University School of Medicine, New Haven, Connecticut, United States; 6Macquarie