1 University of Ottawa, Ottawa, Ontario, Canada; dgoul082@uottawa.ca.
2 Université Laval, Quebec City, Canada; michel.boivin@psy.ulaval.ca.
3 University of Ottawa, Ottawa, Canada.
4 McGill University, Department of Epidemiology, Biostatistics and Occupational Health, Montreal, Canada.
5 University of Ottawa, Department of Mathematics and Statistics, Ottawa, Canada; cgravel@uottawa.ca.
6 University of Ottawa, Epidemiology & Community Medicine, Ottawa, Ontario, Canada.
7 Canada Research Chair in Human Genome Epidemiology, Ottawa, Ontario, Canada; jlittle@uottawa.ca.
8 Université de Montréal, Montreal, Canada; isabelle.ouellet-morin@umontreal.ca.
9 Université Concordia, Département de psychologie, Montreal, Canada; JP.Gouin@concordia.ca.
10 University of Ottawa, Ottawa, Canada; ldubois@uottawa.ca.

The lack of polygenic scores (PGSs) developed for body-mass index (BMI) in children may be problematic because the genetic architecture characterizing BMI changes throughout life. This study aims to describe the genetic susceptibility to obesity in children and to compare two PGSs based on data from adults and children and their association with BMI and discrimination of obesity. The study sample comprises 717 participants aged 4 to 13. Adult- and child-based PGS were evaluated by examining 1) mean BMI across polygenic score risk categories, 2) the capacity to identify obesity with logistic regression and 3) the linear association with BMI z-scores using linear regression. Increases in one standardized unit of adult-based PGS were related to a stronger increase in BMI z-score (ß=0.24-0.39) than PGS derived in children (ß=0.21-0.30). The association between obesity and the child score was higher (OR = 1.75-2.33) than for the adult score (OR = 1.74-2.06) for the age group 4-7 years. The inverse was observed for the age group 8-13 years (ORchild 1.56-1.79 vs. ORadult 1.78-2.54). Both adult- and child-based PGS show strong associations with BMI and risk of obesity, with the adult-based score standing out from 8 years old.