Polygenic Penetrance on Life Course BMI Across Cohorts

• In Event: 1-180 - Genetic Effects on Health and Education Across the Life Course

Integrative Statement

Introduction: Genome wide association studies rely on increasingly large sample sizes in order to have sufficient power for detecting effect sizes of SNPs. While an important advance, this approach suffers from the limitation that such large samples combine individuals across ages, cohorts, and historical periods. Indeed, individual growth trajectories accelerate during specific developmental stages (Frongillo & Lampl, 2011; North et al., 2010; Oken & Gillman, 2003). However, existing studies do not incorporate recent GWAS findings on BMI, and it is unknown how this unfolds across the entire life course. Furthermore, there is existing evidence that BMI shows increasing polygenic penetrance across aged birth cohorts from 1919-1955 (Conley et al., 2016; Liu & Guo, 2015). However, more recent cohorts have faced far more obesogenic environmental conditions for greater periods of their lives. It is unknown whether this trend may have reversed.

Hypotheses: The objective of this study is to examine the genetic contributions to BMI across the life course, covering ages 0 – 88, and cohorts, covering 1924 – 2000. We address two research questions. First, does the association of polygenic scores with BMI vary by age? And second, does this association vary by cohort, as the nutritional environment transitions from scarcity, to abundance, to obesogenic?

Study Population: To address this research objective, we use data from three nationally representative longitudinal studies: the Fragile Families and Child Wellbeing Study (FFCW), the National Longitudinal Study of Adolescent to Adult Health (Add Health), and the Health and Retirement Study (HRS). Together these data cover the life course from birth to 70, and birth cohorts from 1924 to 2000.

Methods: BMI is constructed from either self-report or measured height and weight. We construct polygenic scores for BMI for all respondents using a harmonized set of directly genotyped SNPs across studies and GWAS weights from the GIANT Consortium (Locke et al., 2015). We use the full set of SNPs (p-value cutoff of 1), with no pruning or clumping (Ware et al., 2017). Analyses are restricted to individuals of European Ancestry. We estimate random coefficient growth curve models to assess whether the polygenic penetrance on BMI changes across age. Models are stratified by birth cohort, and the standardized, residualized polygenic score is regressed on the intercepts and linear age slopes.

Results: We find that polygenic penetrance increases with age within each cohort, particularly within the Add Health cohort. For Add Health respondents with a polygenic score one standard deviation above the mean, BMI is predicted to increase an additional .06 per year. The results provide preliminary evidence that polygenic association with BMI increases with age among American young adults in an obesogenic cohort. While polygenic penetrance increased across the HRS cohorts, it declined among the Add Health and Fragile Families cohorts. The results provide suggestive evidence that polygenic penetrance is low in nutritionally scarce and obesogenic nutritional environments.

Authors

• Lauren Gaydosh, University of North Carolina, Chapel Hill
  Presenting Author

• Kathleen Mullan Harris, University of North Carolina at Chapel Hill
  Non-Presenting Author

• Colter Mitchell, University of Michigan
  Non-Presenting Author

• Lauren Schmitz, University of Michigan, Ann Arbor
  Non-Presenting Author

• Erin B Ware, University of Michigan, Ann Arbor
  Non-Presenting Author

• Jessica Faul, University of Michigan, Ann Arbor
  Non-Presenting Author

• Sara McLanahan, Princeton University
  Non-Presenting Author

• Daniel Notterman, Princeton University
  Non-Presenting Author