Identifying Hidden Risk: Metabolomic Patterns in Childhood Obesity

Childhood obesity’s prevalence continues to rise globally, but the biological pathways linking pediatric obesity to early metabolic abnormalities aren’t fully understood. Investigators from Sun Yat-sen University and the Guangzhou Center for Disease Control and Prevention in China aimed to fill this gap through a detailed metabolomic analysis, using machine learning to uncover specific serum biomarkers that may underlie obesity-related metabolic dysfunction in youth.

Study Design and Analytic Framework

The study included 246 children aged 9 to 18 years, evenly split between those with obesity and normal-weight controls matched for age and sex. Researchers profiled 934 serum metabolites, employing a dual machine learning strategy—Least Absolute Shrinkage and Selection Operator (LASSO) regression and random forest with recursive feature elimination (RF-RFE)—to identify obesity-related metabolites independent of metabolic confounders.

The resulting obesity metabolomics signature (OB-MS) integrated 10 core metabolites and was validated through logistic regression. The model demonstrated a receiver-operating characteristic area under the curve (ROC-AUC) of 0.986 in the test cohort, correctly classifying nearly 90% of cases.

Key Findings: Metabolic Fingerprint of Pediatric Obesity

Children with obesity had significantly higher OB-MS scores than their normal-weight peers, and elevated OB-MS scores were consistently associated with metabolic abnormalities, particularly hypertension and hypertriglyceridemia.

Compared with the lowest quartile of OB-MS, children in the highest quartile had:

• A 2.8-fold higher risk of overall metabolic abnormalities (95% CI, 1.3–6.0)
• A 2.4-fold higher risk of hypertension (95% CI, 1.0–5.8)
• A 9.1-fold higher risk of hypertriglyceridemia (95% CI, 2.0–42.3)

Key metabolites that were independently associated with metabolic abnormalities after adjustment for age, sex, and lifestyle factors included:

• 8-hydroxy-2-deoxyguanosine (8-OHdG), an oxidative stress marker, which showed the strongest positive association (OR 1.87, 95% CI, 1.40–2.50)
• N-acetyl-L-histidine and cortisone, which were inversely associated with metabolic abnormalities

A Framework for Further Study

Although the results limit causal interpretation due to the cross-sectional design and the findings are not widely generalizable due to the single regional cohort, the dual machine-learning design offers a framework for replicating the investigation in other populations. Internal validation, while high in this study, will also be necessary in future examinations to strengthen confidence in the reproducibility of the signature. In addition, interventional and longitudinal studies will be needed to determine whether modulating oxidative stress or glucocorticoid pathways alters metabolic outcomes, and whether the OB-MS accurately forecasts future disease development.

The OB-MS model identifies an oxidative stress– and glucocorticoid-linked metabolomic pattern that may serve as early predictors of cardiometabolic risk before overt clinical disease, potentially identifying pathways to target for early intervention.

Reference
Wan J, Luo S, Zhong W, et al. Obesity metabolomics signature in children: associations with metabolic abnormalities and potential biomarkers. Front Endocrinol (Lausanne). 2025;16:1671613. Published September 25, 2025. doi:10.3389/fendo.2025.1671613