MCI was deﬁned as having a) cognitive symptoms, b) a CDR=0.5 and c) score9 on the NYU delayed paragraph recall. We deﬁned Stable MCI to include all individuals who at 4 years had not progressed to Alzheimer Disease (AD). We analyzed 6-month interval data for 209 subjects in the placebo arm who met criteria for Stable MCI. Longitudinal change was investigated on the MMSE and the AD Study Cooperative-Activities of Daily Living (ADCS-ADL) scale. The analysis was performed on all available measures with mixed-effects longitudinal regression. Age, sex and baseline scores were used as covariates. Results: The baseline mean MMSE and ADCS-ADL were 27.8 (1.9) and 60.3 (6.7) respectively. Over 4 years, the MMSE showed a linear trend (p=0.08) towards improvement with a small mean change of 0.11 (95%CI -0.44 to 0.23) at the end of 4 years. The ADCS-ADL also showed a small change of +0.23 (95% CI, -0.74 to 1.19) at the end of four years with an initial positive linear (p=0.01) and then a negative quadratic term (p=0.002). The 95%CI of the mean score remained stable throughout the 48-month study. The interquar- tile range was 1.0 to -1.0 for the MMSE, 6.5 to -3 for the ACDS-ADL. Older age was associated with poorer scores but gender did not show any association. Conclusions: These ﬁndings underscore that within the cur- rently deﬁned condition of MCI a large subgroup of subjects exists whose cognition and ADL can be documented to be remarkably stable over 4 years despite their proposed high risk of progression to AD/dementia, Further studies are needed to understand Stable MCI by identifying the predictive features and etiology of this condition. P3-147 CUT-SCORES FOR DEMENTIA PREDICTION VARY WITH TIME TO DIAGNOSIS: CLINICAL AND RESEARCH IMPLICATIONS FROM THE EINSTEIN AGING STUDY (EAS) Mindy J. Katz, Cuiling Wang, Molly Zimmerman, Richard B. Lipton, Amy Sanders, Herman Buschke, Joe Verghese, Carol Derby, Albert Einstein College of Medicine, Bronx, NY, USA. Contact e-mail: katz@aecom.yu.edu Background: Dementia is often characterized by a lengthy preclinical course. Cut-scores for predicting future dementia are usually “compara- tive” (a cut-score 1.5 standard deviations below the mean of a normal aging sample) or “diagnostic” (the score that distinguishes dementia from aging). Instead, we propose the use of “predictive” cut-scores based on their ability to predict future dementia over various time intervals. Methods: In the EAS (N=727, mean age=78.4, mean follow-up=3.5 years), memory was evaluated using free recall from the Free and Cued Selective Reminding Test (FR-FCSRT; range 0-48). Dementia was diagnosed by consensus conferences (DSM-IV criteria, incident dementia N=90). We identiﬁed a subgroup (N=322, incident dementia N=33) with four consecutive annual evaluations to establish “diagnostic” as well as “predictive” cut-scores using ROC curves. These cut-scores were then applied prospectively to the entire cohort to predict incident dementia one, two or three years after baseline ﬁxing the sensitivity at 80%. Results: In the subgroup, the FR- FCSRT “diagnostic” cut-score was 23 (speciﬁcity-92%) at the time of diagnosis. At one, two and three years prior to diagnosis, FR-FCSRT “predictive” cuts for 80% sensitivity were 28, 32, and 34 with speciﬁcities of 71%, 53%, and 31%. In the overall cohort, a baseline “diagnostic” cut-score of 23 detected only 61% of subjects who demented within three years, while the “predictive” cut-score of 34 detected 84%. The odds ratio of incident dementia over 3 years for FR-FCSRT  23 vs. 28 was 26.6 (95% CI 13.4-53.1). For FR-FCSRT 24-27 vs. 28 the odds ratio of dementia within 3 years remained substantial, 4.9 (95% CI 2.2-10.6). Conclusions: “Predictive cut-scores” improve detection of high risk sub- jects who develop dementia. This approach allows clinicians to interpret an individual’s performance in terms of future dementia risk. When applied to clinical trials, these results highlight the trade-off between restrictive cut- scores which have higher conversion rates to dementia but miss many future cases and less restrictive cut-scores which capture a higher propor- tion of future dementia cases at the price of lower conversion rates. Predictive cut-scores should be used to optimize the design of secondary prevention trials. P3-148 VISUAL MOTION DEFICITS IN ALZHEIMER’S DISEASE DO NOT REFLECT THE FAILURE OF TEMPORAL INTEGRATION Voyko Kavcic, Anthony Monacelli, Roberto Fernandez, William Vaughn, Charles Duffy, University of Rochester Medical Center, Rochester, NY, USA. Contact e-mail: voyko_kavcic@urmc.rochester.edu Background: Previously, we have described deﬁcits in the analysis of patterned visual motion in early Alzheimer’s disease (AD). Most recently, we found that scalp-recorded visual motion evoked potentials (VMEPs) can detect and quantify these deﬁcits. We now explore the role played by temporal integration in these processes, testing the hypothesis that the impaired temporal integration of visual motion underlies these deﬁcits. Methods: We compared perceptual performance and VMEPs in young normal controls (YNC, N = 12), older adult controls (OAC, N = 15), and early AD patients (EAD, N = 13). Visual motion stimuli were presented on a large (60 o X 60 o ) screen display during centered visual ﬁxation. The duration of up- or downward planar motion was varied from 33, 83, 200, to 500 ms as subjects performed a 2AFC discrimination task. Results: Behavioral performance (d’) revealed a main effect of subject group: YNC = OAC  EAD. There was also a strong effect of duration 33 ms  83 ms  200 ms  500 ms. There was no evidence of group by duration interactions, suggesting that AD effects are not linked to temporal duration effects. VMEP amplitudes supported the view that visual motion process- ing deﬁcits in aging and AD are not a reﬂection of temporal integrative failure. The N200 component of the VMEP, most evident at the CPZ electrode, varied between groups in a manner that highlighted the differ- ences between aging and AD effects: YNC = OAC  EAD. The EAD patients also showed a 25 ms delay in the N200 latencies. However, all groups showed parallel increases in N200 amplitudes with increasing stimulus duration in the pattern of 33 ms  83 ms  200 ms  500 ms. Conclusions: We concluded that visual motion perceptual deﬁcits, and corresponding changes in neurophysiological responses, do not reﬂect failures of visual motion integration. Our ﬁndings highlight the utility of using patterned visual motion in psychophysical and neurophysiological testing for the early detection of AD. This approach can successfully distinguishing between cognitive aging and AD and deﬁne speciﬁc patterns of functional decline to guide behavioral and pharmacological interven- tions. P3-149 QUALITY OF LIFE IN ALZHEIMER’S DISEASE AND ATTENTION IN PATIENTS WITH DEMENTIA: A MULTICENTER CROSS- SECTIONAL STUDY Hee-Jin Kim 1 , Hyunseok Kang 2 , Seol-Heui Han 2 , 1 HanYang University Hospital, Seoul, Republic of Korea; 2 Konkuk University Hospital, Seoul, Republic of Korea. Contact e-mail: ewhabrain@naver.com Background: Evaluating quality of life (QoL) in people with dementia has become increasingly valued, for example in assessing the effectiveness of interventions or making treatment decisions, however little is known about the relationship between quality of life and attention of patients with dementia. The aim of this investigation is two-fold, ﬁrst, whether the The Quality of life in Alzheimer’s disease (QoL-AD) and attention question- naire could be used as valuable tools to assess cognitive dysfunction in the elderly. Second, to see if there are any clinically relevant relationships between QoL-AD and sub-items of attention questionnaire, which faith- fully reﬂect the state of attention. Methods: We tried to address these questions using newly developed attention questionnaire and the QoL-AD. Fifty cognitively normal elderly and 218 patients ranging from mild cog- nitive impairment to moderately severe dementia and their caregivers were invited to participate from 15 centers. Results: High QoL-AD composite T563 Poster Presentations P3: