Pleiotropic Approach Identifies Genetic Regions Missed in Single Phenotype Analyses: Example in Alzheimer's Disease (AD) Pathology

• Activity Number: 139
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Lori B. Chibnik*+ and Charles C. White and Towfique Raj and David A. Bennett and Philip L. De Jager
• Companies: Brigham & Women's Hospital/Harvard Medical School and Brigham & Women's Hospital and Brigham & Women's Hospital and Rush Alzheimer's Disease Center and Brigham & Women's Hospital
• Keywords: pleiotropy ; Genetics ; Alzheimer's ; pathology ; ordinal regression ; neurology

Abstract

The standard univariate GWAS lacks power to identify genetic variants with small effects. By applying a pleiotropic approach, we leveraged the shared genetic architecture of pathologically driven cognitive decline to increase power to find associated variants. Using Multiphen, we simultaneously analyzed 7 pathology traits with an inverted ordinal logistic regression model genome wide, where alleles are the outcome and traits are the independent variables. Variants with p-values < 5x10-6 were further investigated in univariate analyses to elucidate traits driving the effect. The top variant is located at APOE (rs429358, p=6x10-29) driven by 5 of 7 traits. Five novel loci were identified driven by 2 to 6 traits. Two SNPs near or in genes are: rs324540 (p=1x10-7, PTPRD), driven by neurofibrillary tangles (p=4x10-6), neuritic plaque (p< 0.04), and macro strokes (p=0.07); and rs12597858 (p=2x10-6, HS3ST4), driven by 6 of 7 traits (3x10-4< p< 0.05). In sum, we discovered several genetic regions that show evidence of pleiotropic association with the pathological traits, which otherwise would have been missed with a univariate approach.