The effect of family history and polygenic risk scores on general and abdominal obesity in the Lifelines Cohort Study

BackgroundPolygenic risk scores (PRSs) have recently become popular in predicting genetic susceptibility to common complex diseases. However, as PRSs only capture part of the genetic risk, imperfect prediction of outcomes is expected. Family history may capture additional genetic effects as well as those of shared family environment. Within-family design have the advantage to control potential bias, such as gene-environmental correlation, assortative mating, and population stratification. Our aims are to investigate the relative influence of PRS and family history on general and abdominal obesity using between and within family approaches. MethodsWe included 50,747 participants of the Lifelines Cohort Study. Body mass index (BMI) and waist-hip-ratio (WHR) were calculated by measured height, weight, waist circumference (WC) and hip circumference. Family history was defined as the number of obese first-degree relatives. PRSs were calculated based on recent large genome-wide association studies for BMI, WC and WHR. The between-family PRS was defined as the mean sibling PRS, and the within-family PRS as the difference between the individual and mean PRS. Linear mixed regression models were used to estimate the effect of (between- and within-family) PRS and family history on BMI, WC and WHR. Structural equation modeling was used to estimate the relative effects of parental PRS (via offsprings PRS) and parental phenotype on offsprings phenotype. ResultsPRS and family history together explained 11.81%, 7.26% and 3.62% of the variance in BMI, WC and WHR, respectively, where family history explained additional variance on top of PRS. Among 2,003 parent-offspring trios, transmission of the PRSs from parents to offspring ({beta}=0.48-0.52) resulted in a significant effect of offsprings PRSs on offsprings BMI, WC and WHR ({beta}=0.14-0.16). In addition, parental BMI, WC and WHR, were strongly positively associated with offsprings BMI, WC and WHR ({beta}=0.14-0.27), independent of the parent-offspring transmission. Between- and within-family PRS had similar effect sizes, while between-family PRS explained more variance than within-family PRS for obesity indices among siblings (1.91% vs 6.70% for BMI, 1.25% vs 4.10% for WC, and 0.72% vs 2.61% for WHR). ConclusionsThe combination of PRS and family history improved the prediction of indices for general and abdominal obesity. In addition to the genetic effects captured by the parent-offspring PRS transmission, parental BMI, WC and WHR also independently influenced offsprings BMI, WC and WHR, reflecting additional genetic effects not captured by PRS as well as effects of the shared family environment. Similar effects of between- and within-family PRSs indicated little bias (e.g., gene-environmental correlations, assortative mating, and population stratification) in the genetic effects on indices of obesity.