was estimated for an 18 month, two-arm, equal allocation trial with 90% power. A neuroimaging profile indicative of pre-clinical AD was determined by a linear classifier trained on discriminating patients diagnosed with AD from normal controls subjects. Results: A trial using the recruitment criteria of the ADNI study would expect a mean placebo arm rate of decline of 0.6 MMSE points/year and require a sample size of 1000 subjects/arm to detect a 50% reduction in mean rate of decline. A trial restricting to subjects with an ApoE E4 allele would expect a mean rate of decline 1.1 points/year and require 395 subjects/ arm to detect a 50% reduction in mean decline. A trial restricting to subjects with a neuroimaging profile consistent with pre-clinical AD would expect a mean rate of decline 1.4 points/year and require 261 subjects/arm. Conclusions: Enrichment strategies can improve the ef- ficiency of secondary prevention trials. Statistical considerations of power and sample size are relevant to discussions of the relative merit of different enhancement strategies. P1-189 CONVERSION TO DEMENTIA OVER A FIVE- YEAR PERIOD AMONG PATIENTS WITH MILD COGNITIVE IMPAIRMENT IN A POLISH FOLLOW-UP STUDY Tomasz Gabryelewicz, Maria Styczynska, Anna Barczak, Boguslaw Wasiak, Anna Pfeffer, Elzbieta Luczywek, Maria Barcikowska, Polish Academy of Sciences, Warsaw, Poland. Contact e-mail: gabryelewicz@cmdik.pan.pl Background: Mild cognitive impairment (MCI) patients can develop demonstrable cognitive impairment, without crossing the threshold for dementia. They have an increased risk of developing dementia, however not all convert to dementia, and some can revert to normal. Methods: The longitudinal study consisted 105 individuals with the diagnosis of MCI according to Peterson’s et al. criteria. All subjects received annual clinical and psychometric examinations for up to mean 5 years. Diag- nosis were made for dementia according to DSM III-R. All patients were given routine laboratory tests, APOE genotype, CT, (1) H MRS, SPECT and standard neuropsychological examinations. Results: 42 subjects were classified as amnestic MCI and 63 as multi-domain MCI. Conversion to dementia over the average of 5 years period was deter- mined. Over the follow-up period 35.3% (37/105) showed cognitive decline not dementia, 27.6% (29/105) were stabile, while 7.6% patients (8/105) showed cognitive improvement. Thirty-one out of 105 patients with MCI (29.5%) converted to dementia: 84% were diagnosed as Alzheimer’s disease (AD), 3% as fronto-temporal dementia, 13% as mixed AD/vascular dementia. Patients with multi-domain MCI were more likely to convert to dementia than those with amnestic MCI. Conclusions: We conclude that conversion rate from MCI to dementia was 29.5% over a 5 years period. Patients with multi-domain MCI were at higher risk for more rapid conversion to clinical dementia. P1-190 EEG BIS-AD PREDICTED DECLINE IN COGNITIVE PERFORMANCE OF SUBJECTS WITH MILD COGNITIVE IMPAIRMENT Scott D. Greenwald 1 , Chuck Smith 1 , Evian Gordon 2 , Philip Devlin 1 , 1 Aspect Medical Systems, Inc., Norwood, MA, USA; 2 Brain Resource Corporation, Inc., Ultimo, Australia. Contact e-mail: sgreenwald@aspectms.com Background: EEG BIS-AD is an index designed to correlate with severity of dementia and cognitive performance (100: fully intact to 0: no cortical activity). Prior work demonstrated that BIS-AD correlates with metrics of cognitive performance (e.g., ADAScog) and is signif- icantly higher in normal subjects compared to patients with Alzheimer’s Disease (AD). We hypothesized that low BIS-AD values would predict worsening of cognitive performance in subjects with Mild Cognitive Impairment (MCI). Methods: Subjects with MCI enrolled in an ongo- ing longitudinal trial (“Determination of the Most Sensitive EEG Bi- omarkers for the Earliest Identification of MCI”) designed to evaluate EEG and neurocognitive biomarkers of MCI and predictors of conver- sion from MCI to Alzheimer’s Disease. Elderly (age = 65) subjects with memory complaints, abnormal memory function (assessed by delayed recall), and reasonable cognitive function (Clinical Dementia Rating Scores (CDRS) = 0.5 and MMSE = 24) were operationally defined as having MCI (a la the Peterson, et al, 1999 criteria). BIS-AD (rev 0.2) was calculated for each subject from a 2-minute segment of frontotemporal EEG recorded during a vigilant, eyes-closed period. Group differences were evaluated using the Student’s t-test and Chi- Square tests as appropriate. P0.05 was considered statistically signif- icant. Results: Data from18 subjects who completed baseline and 6 month visits were available for this interim analysis (baseline: age: 74.9  5.6 ; ADAScog: 12.0  4.4 ; BIS-AD: 97.7  2.2). Subjects were divided in 2 groups using the median split of the baseline BIS-AD values: 99.4  0.4 vs. 96.0  1.9, p0.001. Subjects in the lower BIS-AD group had greater worsening of ADAScog six months later compared to subjects in the higher BIS-AD group: 5.3  5.6 vs. -1.1  5.5, p=0.022. Lower values of baseline BIS-AD predicted worsening of ADAScog with 78% accuracy, 73% Sensitivity and 86% Specificity (p=0.025.) Conclusions: Lower BIS-AD values at baseline in MCI subjects predicted worsening of ADAScog performance 6 months later. Longitudinal follow-up of the MCI subjects will determine whether low BIS-AD values in these subjects are predictive of conversion to Alz- heimer’s Disease. P1-191 BRAIN VOLUME AND VASCULAR RISKS CORRELATE WITH MILD COGNITIVE IMPAIRMENT SUBTYPES Jing He, Sarah Farias, Oliver Martinez, Bruce Reed, Dan Mungas, Charles DeCarli, University of California at Davis, Davis, CA, USA. Contact e-mail: jing.he@ucdmc.ucdavis.edu Background: Mild cognitive impairment (MCI) is a clinically and etiologically heterogeneous syndrome. Although cerebrovascular dis- ease is thought to be etiology for a number of MCI subtypes, few studies have examined the relative impact of vascular disease on MCI subtypes. Methods: Fifty-six amnestic single domain, 37 amnestic multiple domain, 14 non-amnestic multiple domain and 21 non-amnes- tic single domain across an ethnically and racially diverse group of subjects with an average age 75.327.23 years were studied and results compared to 196 cognitively normal (NL) individuals aged 73.1  7.13 years. Hippocampal volume was measured from the 3DT1 image, brain volume and WMH from Fluid Attenuated Inversion Recovery (FLAIR) image. Cerebrovasular risk score was assessed from medical records review. Results: Using multiple linear analyses adjusting for age, gender, education and ethnicity, brain volume (p=0.0034) and hip- pocampal volume (p0.0001) differed significantly between NL and MCI subjects. When comparing the prevalence of MCI subtypes, Af- rican Americans are more likely to have non-memory subtypes as compared to Caucasians and Hispanics. Amongst the MCI subtypes, brain volume varied significantly (p=0.0316). Amnestic MCI had the least brain atrophy, whereas amnestic multiple and non-amnestic mul- tiple had the most brain atrophy. Hippocampal volume did not vary amongst subtypes (p=0.7964). WMH differed significantly between memory and non-memory subtypes (p=0.0397) adjusting for age, gen- der, education and ethnicity. Vascular risk (p=0.035) and lacunar infarcts (p=0.0244) also varied significantly amongst the subtypes. Conclusions: All subtypes MCI show evidence of brain injury by MRI as compared to NL. Hippocampal atrophy was present and did not vary across subtypes, but brain volume varied by subtypes, with multiple domain MCI having the greatest atrophy consistent with more advanced disease. In addition, WMH and lacunar infarcts were higher for non- memory versus memory subtype and vascular risk significantly corre- T264 Poster Presentations P1