Cerebrospinal fluid (CSF) protein concentration (due to defects in the blood- brain barrier), white cell count as well as intrathecal IgG syntesis are poten- tial markers of such processes. The aim of this study was t o investigate whether an elevated CSF protein concentration,an elevated CSF white cell count or an increased IgG synthesis at baseline can predict clinical progres- sion from Mild Cognitive Impairment (MCI) to AD. Furthermore, to inves- tigate the potential role of these markers as risk factors for progression from MCI to AD, in comparison to Ab42, Tau and P-tau levels. Methods: Routine CSF parameters (white cell count and protein concentration) as well as CSF- Ab42, CSF-Total Tau, CSF-P-Tau and IgG index measured at baseline lumbar puncture, were retrospectively registered in a consecutive cohort referred for cognitive evaluation and diagnosed with MCI. Clinical progression was determined based on clinical history and cognitive tests (MMSE) administered at baseline and during follow-up. Results: Fifty-two MCI subjects (21 females, 31 males) were included, mean age 70.3 (range 56 to 80 years), mean MMSE score 27 (range 24 to 30). The patients had a mean follow-up period of 28.5 months (range 5 to 48). Twenty-seven patients remained stable during follow-up (non-progressive MCI), whereas twenty-five patients progressed from MCI to AD (progressive MCI). Neither CSF white cell count nor protein concentration differed significantly between the two groups. None of the patients had an elevated IgG index. No significant difference was found in CSF-Total Tau or CSF-Ab42 concentrations between the groups. The CSF-P-Tau mean concentration was significantly higher in the group of progressive MCI (p ¼ 0.04), although within laboratory reference range. Conclusions: In this retrospective study of CSF parameters as potential baseline predictors of clinical progression in MCI, we found only CSF-P-Tau to be a risk factor for progression. We found no association between CSF protein concentration, CSF white cell count or IgG synthesis and clinical progression of MCI. P4-060 REDUCED CEREBRALVASOREACTIVITY IN DEMENTIA PATIENTS: A MEXICAN COHORT Fernando Gongora-Rivera, Meztli Artemisa Espinosa-Ortega, Antonio Anaya-Escamilla, Humberto Leal-Bailey, Eduardo Adrian Garza-Villarreal, Universidad Autonoma de Nuevo Leon, Monterrey, Mexico. Contact e-mail: meztli.espinosa@hotmail.com Background: New studies have shown that cerebral vascularity may be impaired even in non-vascular dementia, and even that the cause for all types of dementia may be vascular in origin. In this study, we wanted to investigate vascular function in patients with dementia. Methods: We recruited 45 sub- jects, 26 (21 female) Alzheimer’s dementia patients (DM) and 19 healthy controls (15 female) (HC), paired for age and sex. The subjects were re- cruited from the neurology clinic of the Universidad Aut onoma de Nuevo Le on in Monterrey, M exico. We evaluated cerebral vascular function by means of transcraneal doppler ultrasound and vasoreactivity with CO2 chal- lenge, measured in the middle cerebral artery. The variables were cerebral blood flow velocity (CBFV) systolic (CBFV-S), diastolic (CBFV-D) and mean (CBFV-M). The group comparisons were performed using univariate repeated measures ACOVAwith "Phase" (initial - final) as the within-sub- jects contrast, "Group" (DM - HC) as the between-subjects contrast, and "age" and "sex" as covariates. Finally we performed two-tailed non-para- metric correlations between the Mini-Mental (MMSE) questionnaire and the DCBFV-S, DCBFV-D, and DCBFV-M (absolute change represented by D). Results: The descriptive results are shown in Table 1. The statistical analysis showed a significant interaction of Phase x Group for CBFV-S (F (1, 41) ¼ 7.29, p ¼ .01), CBFV-D (F (1, 41) ¼ 6.98, p ¼ .01) and CBFV- M (F (1,41) ¼ 4.54, p ¼ .04) meaning that DM patients had smaller changes in CBFV than healthy controls. We also found positive correlations between MMSE and CBFV-S (r¼ .34, p ¼ .02), CBFV-D (r ¼ .42, p ¼ .004), and CBFV-M (r¼ .30, p ¼ .05). Conclusions: In our study of a Northern Mexican sample, dementia patients showed smaller changes in absolute ce- rebral blood flow velocity with the CO2 test, which were correlated with their cognitive scores. Table 1 Descriptive statistics of the subjects. Healthy Controls Dementia Patients Clinical Age 78 (59 – 90) 78 (67 – 93) Education 6 (0 – 16) 3 (0 – 15) MMSE 27 (63.20) 14.08 (65.80) DM2+ 4 (21%) 9 (35%) HTA+ 8 (42%) 13 (50%) Dyslipidemia+ 9 (47%) 5 (19%) Smoking+ 7 (37%) 6 (23%) Ultrasound CBFV-S Initial 69.90 (618.44) 62.36 (619.83) CBFV-S Final 82.64 (614.47) 68.30 (620.50) CBFV-D Initial 26.36 (67.26) 25.34 (68.21) CBFV-D Final 31.86 (67.42) 27.09 (69.78) CBFV-M Initial 49.00 (615.52) 38.88 (614.95) CBFV-M Final 57.70 (614.08) 43.69 (616.87) DCBFV-S 12.74 (68.27) 5.94 (68.00) DCBFV-D 5.5 (63.61) 1.75 (65.03) DCBFV-M 8.70 (64.14) 4.81 (66.96) + ¼ positive; MMSE ¼ Mini Mental; DM2 ¼ Type 2 Diabetes; HTA ¼ Hypertension. Age, Education ¼ median (min – max); DM2+, HTA+, Dys- lipidemia+, Smoking+ ¼ group frequencies (group percentage). For all other variables ¼ mean (standard deviation). P4-061 LIFESTYLEFACTORS MAY DELAY THE AGE OF CLINICAL DIAGNOSIS OF ALZHEIMER’S DISEASE Rachel Aine Yotter 1 , Xiao Da 2 , Christos Davatzikos 3 , 1 University of Pennsylvania, Section for Biomedical Image Analysis, Philadelphia, Pennsylvania, United States; 2 University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, United States; 3 University of Pennsylvania, Philadelphia, Pennsylvania, United States. Contact e-mail: rachel.yotter@gmail.com Background: Variation in the APOE gene may confer significantly higher risk of developing Alzheimer’s disease (AD). However, having this allele does not guarantee the development of AD, suggesting that - if genetics plays a determining role - it is the interaction between genetics and other factors that results in clinical symptoms (Soininen 2014). To determine other factors that contribute to clinical symptom manifestation, we clustered whole genome sequencing data for apiloprotein genes, then developed a polygenic score to forecast the probable age at which a clinical diagnosis of AD would occur. We then examined lifestyle factors in subjects with delayed onset. Methods: 443 Caucasian participants from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) were selected who had whole genome sequencing data (Table 1). For nine apiloprotein genes, we clustered genetic variation using singular value decomposition. Then for 101 AD subjects, we split the sub- jects into training (n¼51) and testing (n¼50) groups. For the training group, we used 5-fold cross-validated support vector regression with feature selec- tion to predict the age of diagnosis. We then tested the model on the testing group, then obtained a polygenic risk score for non-AD subjects. We tested relationships to the AD-like brain atrophy (Fan 2007), as well as to CSF tau and Ab 42. For lifestyle factors, we examined relationships to BMI, years of education, and blood pressure. In non-AD subjects, we examined cognitive performance using the UW scores for memory and executive function (Gibbons 2012). Since genetic data is non-normal, we used the Spearman correlation coefficient and the Wilcoxon ranksum test for statistical analyses. All factors were adjusted for age and gender before analysis. Results: With variation from the whole APOE gene, we achieved a correlation of 0.366 for subjects in the test set. In the training set, the cross-validated correlation was 0.249. All other gene clusters were not predictive of year of diagnosis. The predicted age of onset in the AD subjects was significantly lower than in con- trol subjects (AD: 75.8 6 3.8s.d.; non-AD: 76.8 6 3.8s.d.; p ¼ 0.03). A higher polygenic risk score was significantly related to more AD-like Poster Presentations: P4 P804