Metabolomic profiles and childhood obesity

Abstract

Objective: To identify metabolite patterns associated with childhood obesity, to examine relations of these patterns with measures of adiposity and cardiometabolic risk, and to evaluate associations with maternal peripartum characteristics.

Methods: Untargeted metabolomic profiling was used to quantify metabolites in plasma of 262 children (6-10 years). Principal components analysis was used to consolidate 345 metabolites into 18 factors and identified two that differed between obese (BMI ≥ 95‰; n = 84) and lean children (BMI < 85‰; n = 150). The relations of these factors with adiposity (fat mass, BMI, skinfold thicknesses) and cardiometabolic biomarkers (HOMA-IR, triglycerides, leptin, adiponectin, hsCRP, IL-6) using multivariable linear regression was then investigated. Finally, the associations of maternal prepregnancy obesity, gestational weight gain, and gestational glucose tolerance with the offspring metabolite patterns was examined.

Results: A branched-chain amino acid (BCAA)-related pattern and an androgen hormone pattern were higher in obese vs. lean children. Both patterns were associated with adiposity and worse cardiometabolic profiles. For example, each increment in the BCAA and androgen pattern scores corresponded with 6% (95% CI: 1, 13%) higher HOMA-IR. Children of obese mothers had 0.61 (0.13, 1.08) higher BCAA score than their counterparts.

Conclusions: BCAA and androgen metabolites were associated with adiposity and cardiometabolic risk during mid-childhood. Maternal obesity may contribute to altered offspring BCAA metabolism.

Keywords: DHEA-S; androgen; branched-chain amino acids; insulin resistance; metabolomics; obesity; steroid hormones; untargeted.

Conflict of interest statement

Conflicts of interest: Steven Watkins and Ryan Michalek are employees of Metabolon Inc.

© 2014 The Obesity Society.

Figures

Figure 1

Comparison of 18 metabolite factor scores between obese (BMI ≥95th %ile) and lean (BMI<85th %ile) childrena a Estimates are adjusted for child sex, age, and race/ethnicity. b 95% CIs reflect Bonferroni’s correction for multiple comparisons (α = 0.05/18).

Figure 2

Associations between Factor 4 and 9 scores and cardiometabolic risk biomarkersa a Estimates are adjusted for maternal education and child age, sex, race/ethnicity, and fast-food intake. b Natural log transformed due to non-normal distributions.