Note: for non-commercial purposes only Metabolomic profiles of childhood obesity Wei Perng, Matthew W. Gillman, Abby F. Fleisch, Ryan D. Michalek, Steven Watkins, Elvira Isganaitis, Mary-Elizabeth Patti, and Emily Oken Obesity Prevention Program Department of Population Medicine Harvard Medical School & Harvard Pilgrim Health Care Institute Boston, MA, USA wei.perng@gmail.com 1
Background The role of metabolomics?? Disturbed Classic metabolite biomarkers profiles of risk Insulin resistance Obesity Inflammation Leptin, Adiponectin HDL, LDL, TG Disease Type 2 DM CVD MetS 2
Research aims Aim 1 Obese vs. lean Difference in metabolite profiles? Aim 2 Metabolite profiles Classic biomarkers HOMA-IR leptin, adiponectin triglycerides CRP, IL-6 Aim 3 Maternal peripartum characteristics obesity excessive GWG gestational diabetes Child metabolite profile 3
Study population P R O J E C T Viva A Study of Health for the Next Generation 2128 live singleton births Mid-childhood 648 eligible 262 mother-child pairs Criteria: Fasting blood from child Maternal characteristics Target sample size: 300 Include: all child obesity & GDM Criteria: adequate serum volume 4
Study population Characteristics of 262 Project Viva mother-child pairs Mean ± SD or % (N) Child Child's age (years) 8.0 ± 0.9 Female 50% (131) Obesity (BMI 95 th percentile) 32.1% (84) Race/ethnicity White 56.3% (147) African American 24.1% (63) Hispanic 6.9 %(18) Mother Mother's age at enrollment (years) 32.0 ± 5.8 Pre-pregnancy obesity (BMI >30 kg/m 2 ) 22.5% (59) Excessive gestational weight gain 58.8% (154) Gestational diabetes 8.4% (22) 5
Methods Serum (n = 262) Step 1: Untargeted metabolomic 345 metabolites profiling Step 2: Principal components analysis (PCA) Step 3: Compare factor scores Obese (BMI 95 th %ile) Lean (BMI<85 th %ile) Step 4: Multivariable linear regression 18 factors VS. Metabolite factors scores Classic biomarkers n = 84 n = 150 Maternal peripartum characteristics Metabolite factor scores 6
Results Difference (95% CI) in factor scores* 2,0 1,5 1,0 0,5 0,0-0,5-1,0 Aim 1: Comparison of factor scores between obese & lean children Factor 4: Branched-chain amino acids (BCAA) Valine Leucine Phenylalanine Isoleucine Proprionylcarnitine (C3) 2-methylbutyrylcarnitine (C5) Isovalerylcarnitine Isobutyrylcarnitine Tryptophan 3-methyl-2-oxovalerate Kynurenine Tyrosine Gamma-glutamylleucine 4-methyl-2-oxopentanoate Factor 9: Androgen hormones 4-androsten-3beta,17beta-diol disulfate 1 DHEA-S Epiandrosterone sulfate Androsterone sulfate 4-androsten-3beta,17beta-diol disulfate 2 Pregn steroid monosulfate Pregnen-diol disulfate Pregnenolone sulfate Andro steroid monosulfate 2-1,5 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15 F16 F17 *Estimates are adjusted for child s sex, age, and race/ethnicity; 95% CIs reflect Bonferroni s correction. F18 7
Results Difference (95% CI) in biomarker* 0,50 0,40 0,30 0,20 0,10 0,00-0,10 Aim 2: Associations of BCAA & androgen metabolite patterns with cardiometabolic risk biomarkers BCAA-related pattern Androgen pattern -0,20 HOMA-IR Triglycerides (mg/dl) Leptin (ng/ml) Adiponectin (µg/ml) CRP (mg/l) IL-6 (pg/ml) *Adjusted for maternal education and child age, sex, race/ethnicity, and fast-food intake. All biomarkers are ln-transformed. 8
Results Difference (95% CI) in factor score* 1,20 0,70 0,20-0,30-0,80 Aim 3: Associations of maternal peripartum characteristics with offspring metabolite patterns + child BMI BCAA-related pattern Androgen pattern -1,30 Pre-pregnancy obesity Excessive gestational weight gain Gestational diabetes mellitus *Adjusted for maternal education and child age, sex, race/ethnicity, and fast-food intake. 9
Conclusions Aim 1 Childhood obesity BCAA Androgen hormones Aim 2 BCAA Androgen hormones Worse cardiometabolic profile Aim 3 Maternal obesity BCAA in offspring 10
Acknowledgements Dr. Emily Oken Funding/Support: This study was funded by the US NIH (K24 HD069408, R37 HD 034568, P30 DK092924). Sheryl L. Rifas-Shiman for data management and preparation. The mothers and children of Project Viva. Past and present Project Viva staff. 11
Questions? 12
Associations of maternal peripartum characteristics with offspring metabolite patterns, accounting for child BMI Difference (95% CI) in factor score* 1,20 0,70 0,20-0,30-0,80-1,30 Pre-pregnancy obesity Excessive gestational weight gain Gestational diabetes mellitus *Adjusted for maternal education and child age, sex, race/ethnicity, fast-food intake, and child s BMI z-score 13
Factor 4: Plasma metabolite concentrations of obese versus lean children Metabolite concentration z-score Obese Lean Factor BMI 95 th %ile BMI <85 th %ile loading n = 84 n = 150 Factor 4: mean ± SD = 0.00 ± 1.52 Valine 0.83 1.17 ± 0.30-0.54 ±0.27 <0.0001 Leucine 0.76 1.38 ± 0.32-0.37 ±0.29 <0.0001 Phenylalanine 0.72 1.23 ± 0.33-0.27 ± 0.29 <0.0001 Isoleucine 0.71 1.18 ± 0.31-0.37 ±0.27 <0.0001 Proprionylcarnitine (C3) 0.66 0.95 ± 0.29-0.44 ±0.25 <0.0001 2-methylbutyrylcarnitine (C5) 0.63 0.80 ± 0.29-0.03 ±0.26 0.0002 Isovalerylcarnitine 0.60 1.34 ± 0.30-0.33 ±0.26 <0.0001 Isobutyrylcarnitine 0.56 1.19 ± 0.35 0.33 ± 0.31 0.001 Tryptophan 0.54 1.07 ± 0.33 0.21 ± 0.29 0.0007 3-methyl-2-oxovalerate 0.52 0.92 ± 0.30-0.11 ± 0.26 <0.0001 Kynurenine 0.52 0.56 ± 0.32-0.19 ± 0.28 0.002 Tyrosine 0.51 1.54 ± 0.28 0.04 ± 0.25 <0.0001 Gamma-glutamylleucine 0.51 0.53 ± 0.32-0.23 ± 0.28 0.002 4-methyl-2-oxopentanoate 0.51 0.73 ± 0.30-0.23 ± 0.26 <0.0001 a Estimates represent mean ± SE and are adjusted for child's sex, age, and race/ethnicity. b From a t-test. * Indicates tier 2 identification in which no commercially available authentic standards could be found, however annotated based on accurate mass, spectral and chromatographic similarity to tier 1 identified compounds. P b 14
Factor 9: Plasma metabolite concentrations of obese versus lean children Metabolite concentration z-score Obese Lean Factor BMI 95 th %ile BMI <85 th %ile loading n = 84 n = 150 Factor 9: mean ± SD: 0.00 ± 1.62 4-androsten-3beta,17beta-diol disulfate 1* 0.86 0.61 ± 0.31-0.49 ± 0.27 <0.0001 Dehydroepiandrosterone sulfate (DHEA-S) 0.84 0.45 ± 0.27-0.41 ± 0.23 <0.0001 Epiandrosterone sulfate 0.79 0.14 ± 0.32-0.70 ± 0.29 0.0007 Androsterone sulfate 0.79 0.30 ± 0.29-0.38 ± 0.26 0.002 4-androsten-3beta,17beta-diol disulfate 2* 0.78 0.48 ± 0.35-0.74 ± 0.31 <0.0001 Pregn steroid monosulfate* 0.76 0.70 ± 0.30-0.22 ± 0.26 <0.0001 Pregnen-diol disulfate* 0.70 0.30 ± 0.30-0.14 ± 0.26 0.05 Pregnenolone sulfate 0.65 0.71 ± 0.33-0.09 ± 0.29 0.002 Andro steroid monosulfate 2* 0.61 0.41 ± 0.29-0.46 ± 0.26 <0.0001 a Estimates represent mean ± SE and are adjusted for child's sex, age, and race/ethnicity. b From the Wald chi-squared test. * Indicates tier 2 identification in which no commercially available authentic standards could be found, however annotated based on accurate mass, spectral and chromatographic similarity to tier 1 identified compounds. P b 15