|| Checking for direct PDF access through Ovid
Background: The postnatal period can function as a window of time for metabolic programming. As evidence from human observational studies is scarce, we examined the role of postnatal weight, length (linear growth), and weight-for-length (WFL) growth trajectories as effect modifiers of established single nucleotide polymorphism (SNP) associations on lipid levels in a cohort of adolescents from the Santiago Longitudinal Study (SLS) (n=484).Methods: Growth trajectories were characterized via two different approaches: nonlinear mixed effects model (SITAR) and a latent growth mixture model (LGMM). We assessed gene-environment interaction in an additive model within: 1) SITAR including product terms that accurately reflects the trichotomous genetic term, and 2) LGMM using stratified SNP-lipid associations by latent growth patterns. Bonferroni-corrected significant findings are reported.Results: SITAR models did not reveal any evidence of gene-environment interaction. In contrast, given three LGMM patterns of growth, gene-environment interactions emerge for both weight and WFL trajectories. One group of infants had lower velocity but higher acceleration; another group had medium velocity and lower acceleration; a third group had high velocity and lower acceleration. The association between the rs7412 (APOE) variant and HDL (mg/dL) was negative for the low velocity/high acceleration weight trajectory group (mean= -10.2; 95% CI = -16.0, -4.5; n~23) compared to the high velocity/low acceleration group (mean = 11.3; 95% CI = 2.6, 20; n~152). Similarly, the association (95% CI) between the rs78536982 (BAI3, LMBRD1) variant and triglycerides (log(mg/dL) was lower for the low velocity/high acceleration WFL group (-0.61; 0.84, -0.37; n~24) when compared to both the medium velocity/low acceleration (-0.09; -0.23, 0.05; n~271) and high velocity/low acceleration (0.001; -0.11, 0.11; n~189). In sex-stratified analyses, the high velocity/low acceleration group for males (n=254) had a negative association (95% CI) between the rs11076175 (CETP) variant and LDL (mg/dL) (-10.6; -16.7, -4.5; n~140) versus a positive association for the low velocity/high acceleration (12.6; 2.93, 22.2; n~114).Summary: These results demonstrate potential heterogeneity in the genetic association between lipid loci and adolescent lipid levels across different patterns of growth from 0 to 5 months. Future work to examine the role of infant growth as a causal factor in direct and indirect effects is of interest.