Testing for Genetic Association in Case-Control Studies Incorporating Multivariate Disease Characteristics

• Activity Number: 565 - Data Science in Statistical Genomics: Challenges and Solutions
• Type: Topic Contributed
• Date/Time: Wednesday, August 2, 2017 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Genomics and Genetics
• Author(s): Haoyu Zhang* and Thomas U. Ahearn and Montserrat García-Closas and Nilanjan Chatterjee
• Companies: Johns Hopkins Bloomberg SPH and National Cancer Institute Division of Cancer Epidemiology & Genetics and National Cancer Institute Division of Cancer Epidemiology & Genetics and Johns Hopkins University
• Keywords: GWAS ; Heterogeneity ; Two-Stage Modelling

Abstract

As sample size for genome-wide association studies continues to rise, there is unprecedented opportunity for obtaining new insights to genetic architecture of complex diseases. Many diseases like breast cancer are intrinsically heterogeneous consisting of subtypes that could be defined by various pathologic and molecular disease characteristics. We propose a two-stage modeling framework for modeling genetic association in GWAS of cancers utilizing multivariate tumor characteristics. The framework can be used to test for overall genetic association and evidence of etiologic heterogeneity, overall or by specific tumor characteristics. We propose efficient methods for handling missing tumor characteristics so that all cases, irrespective of whether they have complete tumor characteristics data or not, can efficiently contribute to the analysis. Preliminary applications will be illustrated based on analysis of a large GWAS  (Ncase=96317, Ncontrol=111357) of breast cancer incorporating ER, PR and HER2 status, three clinically relevant tumor characteristics.