P323
Poster Presentations: P1 phosphorylated at threonine 181 (P-tau 181P) were available for 204 patients. The effect of AAO and CSF biomarker profiles on the olfactory receptor CNV was assessed in a univariate analysis of variance. Results: The assay identified a CNV spanning 156 kb including the genes OR4M1, OR4N2, OR4K2, OR4K5 and OR4K. Only multiplications of this gene region where identified in AD patients, ranging from 2 to 5 copies. A significant difference in AAO was found between AD patients with a high (mean AAO 77.268.8 years) and a low olfactory receptor copy number (mean AAO 74.868.9; P¼0.02). Furthermore, CSF levels of Ab 1-42 were associated with a high olfactory receptor copy number in AD patients. Conclusions: Our data imply that AD patients with a low olfactory copy number have an earlier AAO of disease. These findings are in contrast with a previous published report, however we were able to identify an effect of the olfactory receptor CNV. The olfactory receptor cluster might potentially be a modifier locus of AAO in AD. Further studies can elucidate the impact of the olfactory receptor CNV and might play an important role in the early detection of disease. P1-054
LINKAGE ANALYSES OF EXTENDED CARIBBEAN HISPANIC FAMILIES INDICATES NOVEL LOCI ASSOCIATED WITH FAMILIAL LATE-ONSET ALZHEIMER’S DISEASE
Christiane Reitz1, Rong Cheng2, Brian W. Kunkle3, Gary Beecham4, Margaret Ann Pericak-Vance5, Lindsay A. Farrer6, Jonathan Haines7, Gerard D. Schellenberg8, Richard Mayeux9, 1Columbia University, New York, New York, United States; 2Columbia University College of Physicians and Surgeons, New York, New York, United States; 3University of Miami, Hussman Institute for Human Genomics, Miami, Florida, United States; 4 University of Miami, Miami, Florida, United States; 5Vanderbilt University, Nashville, Tennessee, United States; 6Boston University School of Medicine, Boston, Massachusetts, United States; 7Case Western Reserve University, Cleveland, Ohio, United States; 8University of Pennsylvania, Philadelphia, Pennsylvania, United States; 9Columbia University, New York, New York, United States. Contact e-mail:
[email protected] Background: 67 Caribbean Hispanic families multiply affected by lateonset Alzheimer’s disease (LOAD), not clustering for the APOEe4 allele, and not harboring known mutations in APP, PSEN1 or PSEN2 were selected for whole genome sequencing (WGS) through the Alzheimer’s disease Sequencing Project (ADSP). Aiming to identify regions that may potentially contain causative sequence variants and are shared within or across sequenced families we performed extensive parametric and non-parametric linkage analyses. Methods: Using available GWAS data on all 469 genotyped individuals from these 67 families, we performed parametric twopoint affected-only and two-point age-penetrance models for LOAD using MERLIN setting the disease allele frequency at 0.001 and penetrances at 0.01, 0.90, 0.90, followed by non-parametric multipoint linkage analyses. Results: Overall, we identified 20 linkage regions with LOD or HLOD scores exceeding 3.5 in two-point or multipoint analyses. Out of these, three regions (3q13.31, 7q36.3 and 14q12) showed overall LOD or HLOD scores above four. Family-specific linkage and haplotype segregation analyses demonstrated LOD scores > 0.7 and complete segregation in four multiplex families at the 3q13.31 locus, LOD scores > 0.7 and complete segregation at the 7q36.3 locus in two families, and LOD scores > 0.8 and complete segregation at the 14q12 locus in three families. The 14q12 locus is located w30kb upstream of PSEN1 for which these families screened negative. Conclusions: Linkage analyses in these 67 multiplex families not clustering for the APOEe4 allele identify at least three loci that may harbor genetic variants causatively associated with familial LOAD. P1-055
MULTIPLE ABCA7 MISSENSE VARIANTS MINED FROM THE EXOME VARIANT SERVER SHOW INDEPENDENT ASSOCIATION WITH INCREASED OR DECREASED RISK OF LATE-ONSET ALZHEIMER’S DISEASE (LOAD)
Christopher William Medway1, Samer Abdul-Hay2, Tynickwa Mims1, Fanggeng Zou1, Li Ma3, Gina Bisceglio1, Shane Pankratz4, Sigrid Sando5,
Jan Aasly6, Maria Barcikowska7, Joanna Siuda1, Zbigniew Wszolek3, Owen Ross1, Minerva M. Carrasquillo8, Dennis W. Dickson1, Neill R. Graff-Radford3, Ronald Carl Petersen9, Nilufer Ertekin-Taner8, Kevin Morgan10, Steven Younkin11, 1Mayo Clinic, Jacksonville, Florida, United States; 2Mayo Clinic, Atlantic Beach, Florida, United States; 3Mayo Clinic Jacksonville, Jacksonville, Florida, United States; 4Mayo Clinic, Rochester, Minnesota, United States; 5Norwegian University of Science and Technology, Trondheim, Norway; 6NTNU, Trondheim, Norway; 7Polish Academy of Sciences, Warsaw, Poland; 8Mayo Clinic Florida, Jacksonville, Florida, United States; 9Mayo Clinic Rochester, Rochester, Minnesota, United States; 10University of Nottingham, Nottingham, United Kingdom; 11 Mayo Clinic, Jacksonville, Florida, United States. Contact e-mail:
[email protected] Background: Recent genome-wide association studies (GWAS) of LOAD have identified SNPs that show significant association at APOE and 19 additional loci. Among the functional variants at these loci, those that alter proteins are particularly important because they can readily be investigated in model systems to search for novel therapeutic targets. The exome variant server (EVS) catalogs whole exome sequencing of 4300 unrelated European Americans, a series large enough to detect virtually all exonic variants with a minor allele frequency (MAF) of 0.1% (1/1000) or more. Thus it is now possible to perform a meaningful, low-cost search for "actionable" variants with MAF > 0.1% by genotyping protein-altering variants cataloged on the EVS in large European American case-control series. Methods: To explore the utility of this approach, we targeted the GWAS-identified ABCA7 locus. Nine missense ABCA7 variants with MAF > 0.1% mined from the EVS were genotyped in our LOAD case control series of 9089 subjects. Results: Allelic association analysis showed that six of the nine missense variants had significant or suggestive association with LOAD (P¼ 0.0013-0.063). To evaluate the independent contributions of each missense SNP and the GWAS tag SNP (rs3764650), stepwise multivariate logistic regression was performed adjusting for sex, age, and APOE e 4 dosage. Remarkably, this analysis showed that the tag SNP (P¼1.30x10 -05, OR¼1.34(1.171.53)), R1812H (P¼1.60x10 -03, OR¼2.18(1.33-3.55)), and A676T (P¼2.30x10 -03, OR¼0.04(0.00-0.21)) independently associate with LOAD. In addition, there were 4 variants (H395R, H463R, N718T and Q1686R), co-linear because they are linked on one haplotype, that entered the model with suggestive P values ranging from 0.067 - 0.108. Haplotype analysis showed highly significant association overall (Global P¼4.38x10 -07) and confirmed the allelic heterogeneity at this locus. Conclusions: Thus our study (i) identifies multiple ABCA7 missense variants that may profitably be studied to understand how ABCA7 function may be modified to increase or decrease risk of LOAD and (ii) indicates that analysis of protein-altering variants cataloged on the EVS can be a cost-effective way to identify actionable functional variants at recently discovered LOAD loci. Follow-up studies will be important to confirm the significance of the associations reported here and to better define their odds ratios. P1-056
GENOME-WIDE ASSOCIATION STUDY USING ALZHEIMER-RELATED PATHOLOGICAL BIOMARKERS
Cynthia Picard1, Cedric Julien2, Doris Dea1, Louise Theroux1, Justin Miron1, Judes Poirier3, 1Douglas Mental Health University Institute, Montreal, Quebec, Canada; 2Sainte Justine University Hospital Research Center, Montreal, Quebec, Canada; 3Douglas Mental Health University Institute, Montreal, Quebec, Canada. Contact e-mail: cynthia.picard@ mail.mcgill.ca Background: Genome-wide association study (GWAS) is a powerful technique to identify polymorphisms associated with a particular disease. However, standard case-control study can often be disappointing because no SNPs reach genome-wide significance. Another way to highlight potential disease-modifying SNPs is to perform the study with pathology-specific biomarkers. This method was applied for late-onset Alzheimer’s disease (LOAD). Methods: GWAS was performed with Illumina 550-Quad BeadChip using DNA samples from Quebec Founder Population (QFP). A total of 629 French Canadians (384 LOAD patients and 245 age-matched control
