Assessing GxE Using Nuclear Families: Can We Make It Robust?

• Activity Number: 441
• Type: Invited
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Clarice Weinberg*+ and Min Shi and David Umbach
• Companies: National Institute of Environmental Health Sciences and National Institute of Environmental Health Sciences and National Institute of Environmental Health Sciences
• Keywords: case-only ; family-based analysis ; gene-by-environment interaction ; environment

Abstract

Complex diseases arise through joint action of environmental factors and genetic susceptibility variants. Investigators have assumed that the robustness protecting family-based genotype analyses from bias due to population stratification will protect assessments of interaction. Not true. When testing against a multiplicative joint effects model, the case-only design offers good power, but is invalid if G and E are correlated in the source population. Four mechanisms can produce G/E dependence: 1. Subpopulations vary in both ancestry and cultural traditions; 2. Family history influences behavior through avoidance of exposures perceived as risky; 3. G influences E; or 4. Selective attrition produces correlation in older age strata. We propose a sibling-augmented case-only approach, which is robust against the first two mechanisms, and is therefore valid when a young-onset disease is studied and G does not influence E. With this design an unaffected sibling is randomly selected and exposure ascertained for that sibling. The case is genotyped if the case-control pair is exposure-discordant. A logistic model permits testing of the exposure main effect and GxE interaction.