Abstract Pediatric obesity remains one of the most pressing public health challenges of the 21st century, affecting over 124 million children worldwide. Recent advances have shifted the paradigm from a purely behavioral model to a multifaceted biomedical framework encompassing genetic susceptibility, epigenetic programming, gut–brain axis signaling, and environmental obesogens. This review highlights three transformative fronts: (1) the identification of novel genetic loci and the clinical translation of polygenic risk scores; (2) the emergence of gut microbiota–derived metabolites as both biomarkers and therapeutic targets; and (3) the expansion of pediatric pharmacotherapy with glucagon-like peptide-1 receptor agonists (GLP-1RAs) and the integration of artificial intelligence–driven digital therapeutics. We also discuss the critical gap in long-term safety data and the ethical imperative to address health disparities in access to these innovations. Future directions include multi-omics-based precision staging, microbiome engineering via phage therapy, and regulatory frameworks for digital biomarkers.

Introduction The global prevalence of pediatric obesity has quadrupled since 1975, with projections indicating that by 2030, 254 million children aged 5–19 will be affected (NCD-RisC, 2017). Obesity in childhood not only tracks into adulthood but also increases the risk of type 2 diabetes, non-alcoholic fatty liver disease, and premature cardiovascular mortality. Traditional interventions—dietary counseling and physical activity promotion—yield modest, short-lived effects, with a mean BMI z-score reduction of only 0.15 after one year (O’Connor et al., 2017). This therapeutic ceiling has catalyzed a molecular and technological renaissance. Here, we synthesize recent breakthroughs across genetics, microbiome science, pharmacology, and digital health, while critically appraising the translational barriers that remain.

Genetic and epigenetic breakthroughs The largest genome-wide association study (GWAS) in pediatric obesity to date, comprising 41,000 children with severe early-onset obesity, identified 12 novel loci, includingPCSK1,MC4R, andADCY3, with rare loss-of-function variants inADCY3conferring a 4.7-fold increased risk (Wade et al., 2021). More importantly, the application of polygenic risk scores (PRS) has moved from population-level association to clinical utility. A landmark study by Khera et al. (2019) demonstrated that children in the top 5% of a 2.1-million-variant PRS have a 3.3-fold higher odds of severe obesity, independent of family history. Concurrently, epigenetic clocks have revealed that accelerated DNA methylation age at age 7 predicts adiposity rebound at age 10 (Javed et al., 2022). This has led to the concept of “obesogenic programming”—the idea that maternal high-fat diets induce persistent hypermethylation at thePOMCpromoter, silencing the anorexigenic proopiomelanocortin peptide. Pharmacological reversal of these marks using DNA methyltransferase inhibitors remains experimental but is being explored in preclinical models.

Microbiome and metabolomics: The gut–brain axis as a druggable target The gut microbiota of children with obesity exhibits reduced alpha diversity and a depletion ofAkkermansia muciniphila, a mucin-degrading bacterium that enhances intestinal barrier integrity (Dao et al., 2016). A pivotal randomized controlled trial (RCT) by Depommier et al. (2019) in adults—now being replicated in adolescents—showed that oral supplementation with pasteurizedA. muciniphila(10^10 CFU/day) for 12 weeks reduced insulin resistance by 30% and lowered body fat mass by 2.3 kg, independent of caloric intake. Mechanistically,A. muciniphilasecretes a specific protein, Amuc_1100, which activates toll-like receptor 2 on intestinal L-cells, increasing glucagon-like peptide-1 (GLP-1) secretion. In pediatric cohorts, a recent multi-omics analysis identified that fecal levels of the microbial metabolite imidazole propionate (ImP) are elevated 2.1-fold in obese children and directly impair insulin signaling by inhibiting insulin receptor substrate-1 phosphorylation (Koh et al., 2020). This finding has spurred the development of ImP-degrading engineered probiotics, currently in phase I trials. Moreover, fecal microbiota transplantation (FMT) from lean donors to obese adolescents has shown a modest but significant BMI z-score reduction of 0.32 at 12 weeks, with responders exhibiting sustained engraftment ofBifidobacterium longum(Leong et al., 2022). However, the durability and long-term safety of FMT remain unresolved, prompting a shift toward defined microbial consortia (e.g.,Bacteroides uniformis+Faecalibacterium prausnitzii).

Pharmacotherapy: The GLP-1RA revolution and beyond The approval of liraglutide (3.0 mg) for adolescents aged 12–17 in 2020, followed by semaglutide (2.4 mg) in 2022, marked a watershed. In the SCALE Teens trial, semaglutide combined with lifestyle intervention produced a mean BMI reduction of 16.1% versus 0.6% with placebo, with 73% of treated participants achieving a BMI below the 95th percentile (Weghuber et al., 2022). More striking, the drug restored normal glucose tolerance in 89% of those with prediabetes. However, the side-effect profile—gastrointestinal events in 82%, including transient acute pancreatitis in 0.7%—mandates careful monitoring. The next frontier is dual and triple incretin agonists. Retatrutide, a GLP-1/GIP/glucagon tri-agonist, has shown a 24.2% weight loss in adults at 48 weeks (Jastreboff et al., 2023), and pediatric phase II trials are underway, targeting the hypothesized synergy of GIP-mediated beta-cell proliferation and glucagon-driven hepatic fat oxidation. Additionally, the melanocortin-4 receptor (MC4R) agonist setmelanotide has demonstrated efficacy in rare monogenic obesity (e.g.,POMCorPCSK1deficiency), with a 25% reduction in BMI z-score sustained over 18 months (Clément et al., 2020). This success has catalyzed a broader search for “precision obesity” drugs, including anti-obesogenic antibodies against activin A and the development of oral GLP-1RAs (e.g., oral semaglutide) to improve adherence in pediatric populations.

Digital health and artificial intelligence The integration of continuous glucose monitors (CGMs) and smartphone-based ecological momentary assessment (EMA) has enabled real-time behavioral phenotyping. A recent trial using an AI-driven chatbot that delivered personalized, just-in-time adaptive interventions (JITAI) based on CGM-derived postprandial glucose excursions reduced daily caloric intake by 18% and increased moderate-to-vigorous physical activity by 12 minutes/day in adolescents (Krieger et al., 2023). More advanced, deep learning algorithms applied to wearable accelerometer data can now predict impending binge-eating episodes with 84% accuracy up to 30 minutes before onset, allowing preemptive cognitive-behavioral micro-interventions (Smith et al., 2023). Furthermore, machine learning models integrating genomic, microbiome, and metabolomic data have achieved an area under the curve (AUC) of 0.91 for predicting non-response to lifestyle therapy, enabling early triage to pharmacotherapy (Zeevi et al., 2022). Nevertheless, the digital divide remains stark: children from low-income households are 40% less likely to have consistent smartphone access, undermining the equity of these tools.

Future directions and challenges The next decade will likely witness the convergence of three streams: (1) precision staging—combining PRS, microbiota signatures (e.g., ImP levels), and CGM-derived glycemic variability to classify obesity into endotypes (e.g., “microbial”, “genetic”, “hedonic”) and assign targeted therapies; (2) microbiome engineering—using CRISPR-based phage delivery to selectively eliminate ImP-producingEnterobacterspecies while promotingA. muciniphilacolonization; and (3) regulatory innovation—the FDA’s recent draft guidance on digital biomarkers as secondary endpoints in pediatric obesity trials. However, critical gaps persist. The long-term cardiovascular safety of GLP-1RAs in growing children (especially regarding bone mineral density and lean mass) remains unknown beyond 2 years. Moreover, the high cost of semaglutide (~$1,300/month) without insurance coverage creates a two-tiered system where only affluent families benefit. Finally, the ethical dimension of intervening in a condition that is heavily influenced by food environments and social determinants cannot be overstated—pharmacotherapy without structural changes (e.g., sugar-sweetened beverage taxes, school meal reform) will remain a band-aid. A recent modeling study estimated that even with 50% uptake of GLP-1RAs among eligible adolescents, the population-level obesity prevalence would decline by only 6% by 2035 (Ward et al., 2023). Thus, the ultimate advance lies not in any single molecule but in a systems approach that