(p ¼ 0.013-0.095). Figure 1 shows the association between performance ona speeded dexterity task and plasma isoflavone levels. Conclusions: Sixmonths of 100mg/day treatment with soy isoflavones did not benefit cogni-tion in older men and women with Alzheimer’s disease. However, given thevariabilityinolderadults’abilitytometabolizeandabsorboralformsofiso-flavones, we examined the association between changes in plasma levels oftotal isoflavones and changes in cognition. While preliminary, these datasuggest the possibility of a dose response effect, such that only individualsable to convert aglycone forms of isoflavones to the glycosidic metaboliteshave the potential to demonstrate cognitive effects.P4-417THE EFFECTS OF STATIN USE ON ALZHEIMER’SDISEASE CSF BIOMARKERS IN COGNITIVELYHEALTHY ELDERLYLidia Glodzik1, Vanessa Bikhazi2,1New York University, New York, NewYork, United States;2New York University School of Medicine, New York,New York, United States.Background: Treatment with statins has been related to a decreased risk ofAlzheimer’s disease (AD) in epidemiologic studies. Similarly, animal datasuggested a positive influence of statins on amyloid-beta levels in the brainand cerebrospinal fluid (CSF). Previous small treatment trials in humanswhichusedCSFbiomarkersasoutcomemeasureshaveyieldedaconflictingresults: reporting no treatment effects or reduction in phosphorylated tau.There is no information about the effects of statins on CSFAD biomarkersin large, longitudinally followed cohort of healthy individuals. Methods:We compared the concentration of total tau (t-tau), phosphorylated tau(p-tau181) and amyloid beta 42 (abeta 42) in statin users and non-usersfrom among 184 (mean age 62 6 9.6 years, 63% women) cognitivelyhealthy subjects participating in longitudinal studies of aging at the Centerfor Brain Health, New York University School of Medicine. They receiveda thorough medical, neurological, psychiatric, radiological, and laboratoryexaminations and underwent lumbar puncture. Statins use was determinedbased on medical history and chart review. The determination of AD CSFbiomarkers was done using ELISA assays. In 103 subjects all evaluations,including CSF examination, were repeated on average 2.0 6 .60 years later.Results: The use of statins (n ¼ 26 subjects) was associated with signifi-cantly lower t-tau (F4,179 ¼ 8.6, p ¼ .004, after accounting for age, genderand cholesterol levels). After adding ApoE genotype (ApoE4+ vs. ApoE4-)to the model this association was stronger (F5, 131¼ 14.7, p <.0001).Similarly, statins users tended to have lower p-tau181 (F3,101 ¼ 2.9, p ¼.09, after accounting for ApoE genotype). The groups did not differ in theirabeta42levels.Longitudinally,wedidnotfindsignificantdifferencesinbio-markers’rates of change between statins users and non-users. Conclusions:Our results support epidemiologic data which linked statin use to a reducedrisk for AD. They do not confirm the link between statins and amyloid, butrathersuggestalternativepathwaysinvolvingtauandtauphosphorylationasa putative mechanisms of HMG-CoA reductase inhibitors action, in agree-ment with earlier human CSF studies.P4-418A RANDOMIZED, DOUBLE-BLIND, PLACEBO-CONTROLLED 16 WEEK STUDY OF THE H3RECEPTOR ANTAGONIST GSK239512 INSUBJECTS WITH MILD-TO-MODERATEALZHEIMER’S DISEASE (AD)Richard Grove1, Joseph Horrigan2, Conn Harrington3, Isabel Beresford3,Andreas Mahler4,1GlaxoSmithKline, London, United Kingdom;2GlaxoSmithKline, RTP, North Carolina, United States;3GlaxoSmithkline,Stevenage, United Kingdom;4Private Practice, Achim, Germany.Background: GSK239512 is a highly potent, brain penetrant histamine H3receptor antagonist that has shown favourable cognitive effects in a 4-weekpilot study using the Cogstate neuropsychological test battery. Methods: Inthis 16-week, double-blind, randomised, parallel group study, 196 currentlyuntreated subjects with mild-to-moderate (MMSE 16-24) Alzheimer’s dis-ease (AD) received GSK239512 (n ¼ 97); or placebo (n ¼ 99) administeredorally each morning. GSK239512 was up-titrated over 4 weeks in a flexiblemanner(10ug-20ug-40ug-80ug)followedbya12-weekmaintenancephase.The co-primary efficacy endpoints were change from baseline in ExecutiveFunction/Working Memory and Episodic Memory composite scores fromthe Cogstate battery at Week 16 (intention-to-treat [ITT], mixed model re-peated measures [MMRM] analyses with a 10% Type I Error rate). Second-ary endpoints included the ADAS-Cog, Attention Composite, and TotalCogstate scores, Mini-mental State Examination (MMSE), Clinician’s In-terview-Based Impression of Change-plus Caregiver Input (CIBIC+), Dis-ability Assessment for Dementia (DAD), Neuropsychiatric InventoryScale (NPI) and Clinician, Patient & Caregiver Global Impressions ofChange (GIC). Results: Study completion rates were GSK239512, 85%;placebo,88%. Statistical separationfromplacebowasobserved onEpisodicMemory at Week 16 (Effect Size (ES) ¼ 0.35) but not on Executive Func-tion/Working Memory (ES ¼ 0.16). Other Cogstate endpoints (Attention1.000.500.0-0.50-1.00<---- In Favour of Active In Favour of Placebo ---->Treatment Differencep=0.8732, ES=0.05p=0.1130, ES=0.41p=0.1343, ES=0.30p=0.8569, ES=0.04p=0.0133, ES=0.52p=0.0747, ES=0.28p=0.0588, ES=0.48p=0.3694, ES=0.22p=0.0495, ES=0.35p=0.3484, ES=0.24p=0.5135, ES=0.16p=0.3804, ES=0.16ModerateMildAllModerateMildAllModerateMildAllModerateMildAllTotalAttentionEpisodicMemoryExecutiveFunction/WorkingMemoryFigure 1. Plot of Cogstate Endpoint Effect Sizes (& 90% CI) at Week 16Developing Topicse58
