factor for dementia, while others show that low cholesterol may be a risk factor. The discrepant findings may be due to when cholesterol is measured over the life course (i.e., mid vs. late life) or, alternatively, relative to the progression of the underlying disease. Objective(s): To examine the relation- ship between cholesterol and dementia in a population-based cohort of women followed for 32 years. Methods: A total of 1,532 women from five birth cohorts born in 1908 (n=81), 1914 (n=180), 1918 (n=398), 1922 (n=464), and 1930 (n=419) were examined in 1968, 1974, 1980, 1992 and 2000. Fasting blood was drawn and cholesterol measured at each examination. Participants underwent a thorough neuropsychiatric examination and dementia was diagnosed by psychiatrists using DSM-IIIR criteria. Cox proportional hazard models were used to examine the relationship between total cholesterol and risk of dementia. Age was used as the time axis, and participants were included in the analysis from the age at which they entered the study until the age at which they developed dementia or age at the last visit or death. All models controlled for blood pressure, body mass index, education and smok- ing. Results: Cholesterol levels rose with age and then began to decline steadily after age 50-60. Over the course of the study, 161 women developed dementia. Risk of dementia was not related to concurrent cholesterol levels (aHR 1.14, 95% CI 0.97-1.33) or to maximum cholesterol levels prior to age 60 (aHR 1.13, 95% CI 0.97-1.32). However, lagging suggested there was an increased risk of dementia with higher cholesterol levels ten years or more prior to onset of disease (aHR 1.21 per mmol/dl unit increase of cholesterol, 95% CI 1.04-1.40). Conclusions: Timing of a cholesterol measurement in relationship to the underlying course of disease may be more important than when it is measured over the life course. High cholesterol ten or more years prior to onset of disease was associated with an increased risk of dementia. This association diminished as the disease progressed towards clinical onset. O2-06-07 RELATION BETWEEN RETINAL MICROVASCULAR SIGNS AND CEREBRAL WHITE MATTER LESIONS: THE AGES- REYKJAVIK STUDY Chengxuan Qiu 1 , Mary Frances Cotch 2 , Siggi Sigurdson 3 , Ronald Klein 4 , Fridbert Jonasson 5 , Palmi V. Jonsson 3 , Olafur Kjartansson 3 , Vilmundur Gudnason 3 , Lenore J. Launer 1 , 1 National Institute on Aging, Bethesda, MD, USA; 2 National Eye Institute, Bethesda, MD, USA; 3 Icelandic Heart Association, Kopavogur, Iceland; 4 University of Wisconsin Medical School, Madison, WI, USA; 5 Medical Faculty, University of Iceland, Reykjavik, Iceland. Contact e-mail: LaunerL@nia.nih.gov Background and Objective: Studies based on middle-aged samples have linked retinal microvascular lesions to clinical stroke and silent cerebrovascu- lar disease. We investigated the relation of subcortical and periventricular white matter lesions (WMLs) to retinal microvascular signs in an elderly population. Methods: The analysis involved 1836 individuals (57.6% women) aged over 65 years who participated in the ongoing population-based Age, Gene/Environment Susceptibility (AGES)-Reykjavik study initiated in 2002. Retinal arteriovenous (A/V) nicking, focal arteriolar narrowing, and microan- eurysms/retinal hemorrhages were evaluated following standardized grading protocols on digital retinal images from both eyes after pharmacologic pupil dilation. Cerebral WMLs were detected with magnetic resonance imaging and WMLs in the subcortical and periventricular regions were rated separately using validated rating scales. Multiple logistic regression models were devel- oped to estimate the odds ratio (OR) and corresponding 95% confidence interval (CI) of retinal microvascular signs associated with subcortical and periventricular WMLs after adjusting for major potential confounders such as smoking, hypertension, diabetes mellitus, and cerebral infarcts. Results: The prevalence of definite retinal microvascular abnormalities in at least one eye were 30.3% for A/V nicking, 13.5% for focal arteriolar narrowing, and 17.6% for microaneurysms/retinal hemorrhages. Overall, A/V nicking was signifi- cantly associated with severity of subcortical and periventricular WMLs. Compared to the lowest quartile of WMLs, the adjusted ORs (95% CIs) of A/V nicking related to the highest quartile of subcortical WMLs were 1.64 (1.21-2.21) and periventricular WMLs were 1.69 (1.22-2.33). The association between A/V nicking and WMLs remained statistically significant or margin- ally significant even when both measures of subcortical and periventricular WMLs were simultaneously entered into the model. Focal arteriolar narrowing and microaneurysms/hemorrhages were associated with heavier load of periventricular WMLs, with adjusted ORs (95% CIs) for highest versus lowest quartile being 1.55 (1.01-2.38) and 1.50 (1.02-2.21), respectively. Focal arte- riolar narrowing and microaneurysms/hemorrhages were not significantly as- sociated with subcortical WMLs. Conclusions: Retinal microvascular lesions are correlated with load of cerebral WMLs in the elderly. This study suggests that, among the elderly population, microangiopathic changes in both retina and the brain may be concomitant. O2-06-08 APOLIPOPROTEIN E PROFILE INFLUENCES BRAIN ACCESS BY HERPES SIMPLEX VIRUS TYPE 1 Javier S. Burgos, Carlos Ramirez, Isabel Sastre, Juan M. Alfaro, Fernando Guzma ´n-Sa ´nchez, Fernando Valdivieso, Centro de Biologı ´a Molecular - Universidad Auto ´noma de Madrid, Cantoblanco, Spain. Contact e-mail: jburgos@cbm.uam.es Background: Alzheimer’s disease (AD) is a devastating neurodegenerative disease of unknown etiology, where genetic and environmental factors may play a role. To date, the most important factors associated to sporadic AD are age, female gender and apolipoprotein E [ApoE]. The risk association of the APOE4 allele and AD has become almost universally accepted. Although the involvement of an infectious agent in the etiology of AD is far from fully demonstrated, Itzhaki et al. have extensively associated the herpes simplex virus type 1 (HSV-1) presence with this disease. The relative risk of develop- ing AD for those positive for HSV-1 DNA in the brain and who carried an APOE4 allele was considerably higher than for individuals with only one (or none) of these factors. Objective: To analyze the effect of ApoE and gender influence on the HSV-1 infectivity to the brain, by using a model of hema- togenous infection of mice. Methods: Wild-type, APOE knockout, APOE hemizygote, APOE3 and APOE4 humanized transgenic mice were used to analyze the influence of the ApoE on the viral loads during acute and latent infection. Moreover, male and female mice were compared. Conclusions: We have found that viral neuroinvasion was reduced in mice lacking ApoE at both acute and latent infection, and that the ApoE dose was directly linked to the HSV-1 cerebral concentration. The ApoE4 animals presented very high levels of virus in the brain in comparison with ApoE3 mice. In the female mice HSV-1 colonization was more effective than in males, especially in the brain. It was also found that ApoE promotes virus colonization of the ovaries, the APOE gene dose being directly related to viral invasiveness. The results presented here indicate that ApoE4 facilitates HSV-1 neuroinvasion and la- tency in the brain much more so than ApoE3, with ApoE dosage correlating directly with the HSV-1 cerebral concentration. Female gender presented a greater brain infection than males. Furthermore, we demonstrated the hema- togenous vertical transmission of HSV-1 from maternal blood to the offspring nervous system, being also APOE-dependent. This group of data strengths the hypothesis that HSV-1 might be involved in AD. TUESDAY, JULY 18, 2006 PLENARY PL3 TUESDAY PLENARY PL-03-01 GRADUAL MEMBRANE CHOLESTEROL REDUCTION AFTER SYNAPTOGENESIS DETERMINES SURVIVAL OF HIPPOCAMPAL NEURONS IN VITRO Carlos Dotti 1 , Federica Chiara 2 , Simona Perga 2 , 1 Department of Human Genetics and Flanders Institute of Biotechnology University Leuven, Leuven, Belgium; 2 Cavalieri Ottolenghi Scientific Institute, University of Turin, Turin, Italy. Contact e-mail: carlos.dotti@med.kuleuven.be S44 Plenary PL3: Tuesday Plenary