Assessment of Rare Variants of Unknown Significance Using Hierarchical Modeling

• Activity Number: 17
• Type: Topic Contributed
• Date/Time: Sunday, July 29, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Colin B Begg*+ and Marinela Capanu
• Companies: Memorial Sloan-Kettering Cancer Center and Memorial Sloan-Kettering Cancer Center

Abstract

Current evidence suggests that the genetic risk of cancer may be attributed primarily to rare variants. Classifying as deleterious or neutral the large number of variants of unknown significance in known susceptibility genes is a difficult on-going task. In this talk we present one approach to this problem, hierarchical statistical modeling using data from case-control studies. We describe our experiences examining the distinctive effects of rare variants in a number of genes that are known to influence cancer. The method can be effective only if higher-level covariates are identified that are observed to be associated with risk. These are then used, implicitly, to group the rare variants into classes with distinctive risks. The higher-level covariates can be bioinformatic characteristics of the variant, or variables that characterize the observed occurrence of cancer in the relatives of all study probands that carry the variant. Our findings from studies of melanoma and breast cancer provide evidence that hierarchical modeling is a useful tool in this context.