Exploring biomarkers related to branched-chain amino acid metabolism in childhood obesity based on transcriptomics and experimental verification

Exploring biomarkers related to branched-chain amino acid metabolism in childhood obesity based on transcriptomics and experimental verification | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Exploring biomarkers related to branched-chain amino acid metabolism in childhood obesity based on transcriptomics and experimental verification Huijie Zhang, Ying Lei, Limei Guan, Hui Liu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9238869/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 8 You are reading this latest preprint version Abstract Childhood obesity (CO) is a complex chronic disease driven by environmental, behavioral, and genetic factors. Increasing evidence suggests that dysregulation of branched-chain amino acid metabolism (BCAAM) contributes to CO development, highlighting the need to identify reliable biomarkers for diagnosis and treatment. In this study, transcriptome data and BCAAM-related genes were obtained from public databases. Differentially expressed genes were intersected with BCAAM-related genes to identify candidates, and machine learning was applied to screen biomarkers and construct a nomogram model. Multi-dimensional analyses, including functional enrichment, immune infiltration, molecular regulatory network, and drug prediction, were further performed. Three biomarkers, PLEK, NIN, and COX1, were identified, and the nomogram based on them showed good predictive performance. These biomarkers were mainly enriched in mitochondrial energy pathways and were significantly associated with multiple differential immune cells. In addition, PLEK and NIN were predicted to interact with several miRNAs, and multiple potential therapeutic drugs targeting these biomarkers were identified. These findings suggest that PLEK, NIN, and COX1 may serve as promising biomarkers and therapeutic targets for CO. Health sciences/Biomarkers Biological sciences/Computational biology and bioinformatics Childhood obesity Branched-chain amino acid metabolism Transcriptomics Biomarkers Immunity Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 1. Introduction Obesity, characterized by abnormal or excessive fat accumulation, is one of the most severe public health problems of this century. Of particular concern is the increasing prevalence in children and adolescents, with global childhood obesity (CO) figures projected to double by 2035 from 2020 levels to 208 million boys (a 100% rise) and 175 million girls (a 125% increase)[ 1 ]. Notably, CO is not only recognized as an independent risk factor for adult obesity but also leads to numerous metabolic or nonmetabolic complications, such as early puberty, type 2 diabetes mellitus (T2DM), non-alcoholic fatty liver disease, cardiovascular diseases, asthma, and even cancer[ 2 ]. Therefore, enhancing early screening and management of CO and thereby enabling timely intervention and treatment through the discovery and validation of novel biomarkers holds significant public health importance for alleviating the future burden of chronic diseases in adulthood. Branched-Chain Amino Acids (BCAAs; valine, leucine, and isoleucine) are essential amino acids that play a pivotal role in lipid, carbohydrate, and protein metabolism. Studies have highlighted a more complex relationship between BCAAs metabolism and the pathophysiology of metabolic diseases, such as obesity, insulin resistance (IR), and type 2 diabetes. Plasma elevations of BCAAs are widely observed in CO, especially when IR is present[ 3 ]. MS-based metabolomics identifies elevated BCAAs as key features of obesity-induced IR, correlating more strongly with IR than lipid metabolites[ 4 ]. Mechanistically, BCAAs drive IR by persistent mTORC1 activation and subsequent IRS-1 phosphorylation, a process reversible by rapamycin in rodent models[ 4 ]. In a mouse model, BCAA supplementation induced obesity and IR, and promoted adipose tissue macrophage M1 polarization via the IFNGR1/JAK1/STAT1 signaling pathway[ 5 ]. The sleep-wake cycle is a critical process for sustaining brain function and overall health, encompassing metabolic and cardiovascular systems. Furthermore, plasma concentrations of BCAAs fluctuate throughout the circadian day. This correlation may be associated with sleep disorders and obesity, and research indicates that BCAAs could be orchestrators between the sleep-wake cycle and metabolism[ 6 ]. Recent research has identified a key function of the BCAA-BCKA axis in white adipose tissue (WAT) browning. Acetyl-CoA derived from branched-chain keto acids (BCKAs) suppresses WAT browning by acetylating PR domain-containing protein 16 (PRDM16) and disrupting its interaction with peroxisome proliferator-activated receptor-γ (PPARγ)[ 7 ]. Furthermore, telmisartan promotes WAT browning and alleviates obesity via BCAT2 suppression[ 7 ]. Research also demonstrated that BCKAs and glutamate/glutamine may be biomarkers of IR in CO[ 8 ]. However, the exact role of BCAAs catabolism in the pathogenesis of obesity is elusive, coupled with a lack of systematic screening and in-depth mechanistic validation of key regulatory genes in pediatric populations. To explore the effect of BCAAs catabolism in CO, we employed a comprehensive strategy by integrating transcriptomics and machine learning to pinpoint key biomarkers, followed by their multi-faceted characterization and experimental validation to assess their diagnostic and therapeutic potential. 2. Results 2.1 Acquisition and functional enrichment analysis of 34 candidate genes To identify genes associated with CO, differential expression analysis was first performed between the CO group and the control group in the training set GSE205668, and 4110 differentially expressed genes (DEGs1) were identified, among which 1795 genes were upregulated and 2315 genes were downregulated in the CO group (Figs. 1 a-b, Supplementary Table S2 ). Subsequently, differential expression analysis was conducted between the CO group and the control group in the training set GSE87493, and 2149 differentially expressed genes (DEGs2) were obtained, with 838 genes upregulated and 1311 genes downregulated in the CO group (Figs. 1 c-d, Supplementary Table S3 ). Genes with consistent expression directions (concomitant upregulation or concomitant downregulation) in both datasets were intersected, ultimately yielding 60 co-upregulated genes and 88 co-downregulated genes, totaling 148 intersecting differentially expressed genes (DEGs3) (Figs. 1 e-f). Subsequently, BCAAM-RGs scores were calculated based on all samples from the training set GSE205668, and results of the Wilcoxon rank-sum test showed that there was a significant difference in BCAAM-RGs scores between the control group and the CO group (P = 0.0026) (Fig. 1 g). No obvious outlier samples were identified by hierarchical clustering analysis of samples (Fig. 1 h ) . Further, WGCNA was applied, and a soft threshold (power) of 12 was selected to construct a scale-free network; at this point, the model goodness of fit R²≈0.8 and the mean value of the adjacency function approached 0, indicating that the network conformed to the scale-free topological characteristics (Fig. 1 i ) . Hierarchical clustering and dynamic tree cutting were performed based on the TOM, with the minimum number of module genes set to 100, and 9 modules were finally obtained (excluding the gray module) (Fig. 1 j ) . Correlations between each module and BCAAM-RGs scores were analyzed, and 6 modules were found to be significantly correlated with BCAAM-RGs scores (|cor|>0.3, P < 0.05). Among them, the brown module showed the strongest negative correlation with BCAAM-RGs scores (cor=-0.83, P < 0.001), and the red module showed the strongest positive correlation (cor = 0.77, P < 0.001). After genes were extracted from these two modules, a total of 998 BCAAM-related key module genes were obtained (Fig. 1 k, Supplementary Tables S4-5 ). After taking the intersection of DEGs3 and BCAAM-related key module genes, a total of 34 candidate genes were obtained (Fig. 1 l, Supplementary Table S6 ). To further explore the biological functions and mechanisms of the candidate genes, GO and KEGG enrichment analyses were conducted on the 34 candidate genes, resulting in 127 GO terms (adj.P < 0.05) (Fig. 1 m, Supplementary Table S7 ), including 93 biological process (BP) terms such as positive regulation of platelet activation and positive regulation of integrin activation; 23 cellular component (CC) terms such as centriolar subdistal appendage and mitochondrial respiratory chain complex III; and 11 molecular function (MF) terms such as microtubule minus-end binding and cytochrome-c oxidase activity. The KEGG enrichment analysis showed that the candidate genes were mainly involved in 12 signaling pathways (adj.P < 0.05), including cardiac muscle contraction, oxidative phosphorylation, and non-alcoholic fatty liver disease, etc (Fig. 1 n, Supplementary Table S8 ). Subsequently, analysis of the interaction relationships between proteins encoded by biomarkers was performed via the STRING database ( https://www.string-db.org ). During data processing, some gene names were automatically converted by the STRING database to internationally accepted mitochondrial gene nomenclature: COX2 was converted to MT-CO2, COX1 to MT-CO1, ND1, ND2, ND3, ND4, and ND5 to MT-ND1, MT-ND2, MT-ND3, MT-ND4, and MT-ND5, respectively, and CYTB to MT-CYB. Results of the analysis showed that close associations existed between mitochondrial respiratory chain components (including MT-ND1, MT-ND2, MT-ND3, MT-ND4, MT-ND5, MT-CO1, MT-CO2, MT-CYB) and PTGS1, suggesting that these proteins may play synergistic roles in energy metabolism processes (Fig. 1 o ) . The candidate genes and enriched pathways identified in this part can provide a theoretical basis for potential therapeutic targets and mechanism studies of CO. 2.2 Identification of 3 biomarkers and construction of a nomogram To further screen reliable biomarkers from 34 candidate genes, LASSO regression analysis, SVM-RFE analysis, and RF analysis were applied in this study. Firstly, LASSO regression analysis identified a feature gene set 1 containing 7 genes at log(lambda.min) = -2.4879, which included EHD4, PLEK, NIN, MAP1B, MYH9, LTB, and COX1 (Figs. 2 a-b ) . Results of SVM-RFE analysis showed that the cross-validation accuracy reached the highest when the number of features was set to 30, and thus a feature gene set 2 containing 30 genes was obtained, including PLEK, PTPN6, MYO9B, ARPC1B, COX1, ALCAM, CYTH4, NIN, etc (Fig. 2 c ) . RF analysis, based on gene importance ranking, selected the top 10 genes in terms of importance from the 34 candidate genes to form feature gene set 3, which included FADD, CTH, ALCAM, NIN, PLEK, PTPN6, ND5, COX1, TBC1D10C, and DOCK2 (Figs. 2 d-e ) . By taking the intersection of the three feature gene sets, 3 biomarkers were obtained, namely PLEK, NIN, and COX1 (Fig. 2 f ) . Subsequently, the nomogram model constructed based on the 3 biomarkers showed good predictive performance. For example, when the total points reached 2.08, the predicted probability of CO was 83.2% (Fig. 2 g). Meanwhile, the performance of the nomogram model was validated using multiple indicators. Results of calibration curve analysis showed that the P value of the Hosmer-Lemeshow (HL) test was 0.85, indicating no significant difference between the predicted values of the model and the actual observed values, with good calibration of the model. In addition, the mean absolute error (MAE) was 0.072, which further confirmed that the error between the actual disease risk and the risk predicted by the model was very small, indicating that the nomogram model had high accuracy in predicting CO (Fig. 2 h). Results of ROC analysis showed that the AUC value was 0.846, demonstrating that the nomogram model had good discriminative ability in distinguishing CO patients from non-CO patients (Fig. 2 i). Furthermore, decision curve analysis (DCA) was used to evaluate the clinical utility of the model, which quantified the net benefit of intervention using the prediction model under different risk probability thresholds. The results indicated that when the high-risk threshold was in the range of 0.1–0.7, the net benefit of the nomogram model curve was significantly higher than that of the negative prediction decision curve (None curve) and the positive prediction decision curve (ALL curve), suggesting that the nomogram model had better predictive performance and application value in clinical practice (Fig. 2 j). This part of the study systematically screened out biomarkers with potential diagnostic value and constructed a reliable nomogram prediction model, providing a theoretical basis for the early, accurate diagnosis of CO and personalized diagnosis and treatment strategies. 2.3 Correlation analysis, functional similarity analysis, and GeneMANIA analysis of biomarkers Correlation analysis of the three biomarkers showed that there were significant correlations between them (|cor| > 0.3, P < 0.05). Among them, the highest positive correlation was observed between NIN and PLEK (cor = 0.65, P < 0.05), and the highest negative correlation was observed between PLEK and COX1 (Fig. 2 D) (cor=-0.73, P < 0.05), suggesting that they may exert a synergistic regulatory effect in the occurrence and development of CO (Fig. 3 a). Results of functional similarity analysis indicated that the average semantic similarity scores of COX1, PLEK, and NIN were all less than 0.5, indicating low functional similarity among the three. Among them, the scores of PLEK and NIN were relatively close (both approximately 0.400), indicating weak functional similarity between the two (Fig. 3 b). To further explore the functional associations of these biomarkers, the GeneMANIA tool was used in this study to analyze their potential interaction network with functionally similar genes (COX1 had been converted to its official gene symbol MT-CO1 before analysis). This network prediction analysis revealed that the biomarkers and other genes were functionally enriched in multiple biological processes, including aerobic electron transport chain, aerobic respiration, and mitochondrial ATP synthesis coupled electron transport (Fig. 3 c). This part of the results was provided as valuable clues for an in-depth understanding of the potential mechanisms of action of these biomarkers in CO. 2.4 Localization analysis and GSEA of biomarkers The results of chromosomal localization analysis showed that PLEK and NIN were localized on chromosomes 2 and 14, respectively (Fig. 4 a), which laid a foundation for studying their genetic regulatory characteristics. Furthermore, since COX1 is mitochondrial DNA, it could not be localized on chromosomes. Results of subcellular localization showed that NIN was mainly localized in the nucleus, while PLEK and COX1 were mainly distributed in the cytoplasm (Fig. 4 b). This indicated that these biomarkers might exert their respective functions in different regions within the cell. To further explore the relevant signaling pathways and biological mechanisms involved in the occurrence and development of CO by these biomarkers, GSEA was performed in this study. It was found in the study that PLEK was mainly enriched in 1480 pathways, including "The transfer of electrons from NADH to ubiquinone that occurs during oxidative phosphorylation" and "The transfer of electrons through a series of electron donors and acceptors, generating energy that is ultimately used for synthesis of ATP", and so on(Fig. 4 c, Supplementary Table S9 ); NIN was mainly enriched in 1403 pathways, including "The transfer of electrons from NADH to ubiquinone that occurs during oxidative phosphorylation" and "The transfer of electrons through a series of electron donors and acceptors, generating energy that is ultimately used for synthesis of ATP", and so on(Fig. 4 d, Supplementary Table S10 ); COX1 was mainly enriched in 1369 pathways, including "The transfer of electrons through a series of electron donors and acceptors, generating energy that is ultimately used for synthesis of ATP" and "The phosphorylation of ADP to ATP that accompanies the oxidation of a metabolite through the operation of the respiratory chain. Oxidation of compounds establishes a proton gradient across the membrane, providing the energy for ATP synthesis", and so on (Fig. 4 e, Supplementary Table S11 ). For these pathways, the |NES| were all greater than 1, FDRs were all less than 0.25, and the P values were all less than 0.05. These results not only clarified the chromosomal localization and subcellular distribution of PLEK, NIN and COX1, but also revealed the biological processes and regulatory pathways in which PLEK, NIN and COX1 were potentially involved. 2.5 Immune infiltration analysis To further analyze the differences in the immune microenvironment between the CO group and the control group, analysis of immune cell infiltration characteristics of the samples was performed in this study. Results showed that the infiltration abundances of 28 immune cells exhibited significant differences between the two groups. Among them, cells such as activated CD4 T cells, activated CD8 T cells, activated dendritic cell, T follicular helper cells, and type 1 T helper cells had lower infiltration levels in the control group, while showing higher infiltration abundances in the CO group (Fig. 5 a). Further analysis using the Wilcoxon rank-sum test identified 18 immune cells with statistically significant differences between the two groups (P < 0.05), which were defined as differential immune cells, including activated CD8 T cells, activated dendritic cells, and T follicular helper cells, etc. (Fig. 5 b). Correlation analysis between differential immune cells revealed that the strongest positive correlation existed between macrophages and myeloid-derived suppressor cells (MDSC) (cor = 0.95, P < 0.05) (Fig. 5 c). Results of correlation analysis between biomarkers and differential immune cells showed that COX1 was negatively correlated with 17 differential immune cells, among which the strongest negative correlation was observed with effector memory CD8 T cell (cor=-0.80, P < 0.05) (Fig. 5 d); NIN was positively correlated with 13 differential immune cells, with the highest positive correlations with gamma delta T cell and regulatory T cell (cor = 0.67, P < 0.05) (Fig. 5 d); PLEK was positively correlated with 15 differential immune cells, among which the strongest positive correlation was with regulatory T cell (cor = 0.90, P < 0.05) (Fig. 5 d). These results not only confirmed the important roles of biomarkers PLEK, NIN, and COX1 in immune regulation and deepened the understanding of the immunopathological mechanism of CO, but also provided a theoretical basis for the future development of therapeutic strategies targeting specific immune cell subsets. 2.6 Molecular regulatory network analysis and drug prediction analysis of biomarkers Potential regulatory mechanisms of the biomarkers were further revealed by molecular regulatory network analysis. Results showed that PLEK and NIN were targeted by 16 and 2 miRNAs, respectively, while no targeting miRNAs were predicted for COX1 and its standard symbol MT-CO. Furthermore, a total of 63 LncRNAs were predicted for the two miRNAs targeting PLEK (hsa-miR-141-3p and hsa-miR-200a-3p) (Fig. 6 a). To explore potential drugs for treating CO, results of drug prediction analysis showed that 15, 12, and 15 potential targeted drugs were predicted for COX1, NIN, and PLEK, respectively (Fig. 6 b, Supplementary Tables S12-14 ). These results were provided as important theoretical bases and resource clues for an in-depth understanding of the molecular regulatory networks (including miRNA and LncRNA levels) of PLEK, NIN, and COX1 in the pathological process of CO, as well as for screening potential therapeutic drugs. 2.7 RT-qPCR Finally, RT-qPCR experiments were used to verify the expression levels of the 3 biomarkers. The results of the RT-qPCR assay indicated that PLEK and COX1 were found to have significant differences between the control group and the CO group (P < 0.05). Compared with the control group, the expression of PLEK in the CO group was significantly increased, while the expression of COX1 in the CO group was significantly decreased (Figs. 7 a-c). This indicated that the expression changes of these two genes were potentially associated with the occurrence and development of CO. 3. Discussion CO is a pathological process with multifactorial causes, which is characterized by an excessive accumulation of body fat and is frequently accompanied by metabolic disorders. Disrupted BCAAs metabolism promotes CO by triggering mitochondrial dysfunction, which adversely affects the sleep cycle, the browning of WAT, and adipocyte-macrophage functionality[ 5 – 7 ]. Impaired mitochondrial BCAA nitrogen flux in brown adipocytes (BAT) can promote IR[ 9 ]. Here, we identified three hub genes as key biomarkers and showed that a nomogram model integrating them offers robust predictive power. A comprehensive investigation—encompassing functional enrichment, localization, regulatory networks, and drug prediction—coupled with RT-qPCR validation, elucidated their regulatory roles. To our knowledge, this is the first study to explore BCAA metabolism-related biomarkers in CO. These findings provide novel insights for the clinical diagnosis and treatment of CO. Combining WGCNA, limma difference analysis, and machine learning, we identified three hub genes as biomarkers of CO, including PLEK, NIN, and COX1. PLEK, also known as P47, is a substrate for protein kinase C in platelets and leukocytes[ 10 ] and is related to various autoimmune and inflammatory diseases[ 11 ]. A recent study demonstrated that PLEK acts as a hub gene via the MAPK and PI3K-Akt signaling pathways in cuprizone-induced demyelination and cognitive impairment in mice[ 12 ]. Over the past decade, researchers have gained a deeper understanding of the PI3K-Akt signaling pathway, which plays a vital role in metabolic diseases[ 13 ]. Modulation of the PI3K-AKT signaling pathway and its downstream molecules represents a potential therapeutic strategy for the treatment of obesity and type 2 diabetes[ 13 ]. The study on chronic periodontitis demonstrated that the p38 MAPK inhibitor significantly reduced pleckstrin levels induced by IL-1β and LPS, suggesting that PLEK may regulate inflammation via the p38 MAPK pathway[ 11 ]. Studies have shown that LPS levels are positively correlated with obesity, and LPS-triggered inflammation enhances the degree of obesity[ 14 , 15 ]. Hypoxic adipocytes recruit and activate immune cells, releasing pro-inflammatory cytokines including IL-1β and TNF-α. The sustained increase in LPS and IL-1β leads to chronic activation of the p38 MAPK pathway across multiple cell types—a key mechanism underlying obesity-related insulin resistance and metabolic syndrome[ 16 – 18 ]. Together, these findings suggest that PLEK may be upregulated by LPS and IL-1β via the p38 MAPK pathway and contributes critically to CO pathogenesis. The mammalian target of rapamycin (mTOR) signaling is associated with numerous cellular processes such as protein synthesis, cellular metabolism, and growth[ 19 ]. Activation of mTOR signaling can reduce food intake[ 20 ]. The study showed that ventricular zone expressed pleckstrin homology domain-containing 1 (VEPH1) acts to inhibit mTORC1 signaling through a mechanism involving enhanced TSC1/TSC2 binding, facilitated translocation of TSC2 to the membrane, and a resultant increase in TSC2 Ser1387 phosphorylation[ 21 ]. It has been demonstrated that leucine induces the dephosphorylation of Sestrin2, leading to its dissociation from GTPase-activating proteins towards Rags 2 (GATOR2) through this mechanism, thereby activating mTORC1 signaling[ 22 – 25 ]. Our findings propose a mechanistic hypothesis whereby leucine-induced mTORC1 activation upregulates PLEK expression, thereby exacerbating inflammation and insulin resistance. Obesity-induced insulin resistance is closely related to adipose tissue inflammation, which is driven by adipose tissue macrophages[ 26 ]. MIN, also called as ninein, plays a crucial role in macrophages phagocytosis[ 27 ]. Branched-chain amino acid transaminase 1 (BCAT1), a key rate-limiting enzyme in BCAA metabolism, is highly expressed in macrophages, and BCAT1-mediated metabolic reprogramming is closely associated with inflammatory activation[ 28 ]. In this study, MIN expression was significantly elevated in CO. Therefore, a hypothese, MIN and BCAAs metabolism regulate the functional status of adipose tissue macrophages, affecting the occurrence and development of chronic inflammation, and thus participating in the pathological process of obesity and its related metabolic complications, maybe reasonable. Research has found that the expression of mitochondrial-related pathway genes is downregulated in the subcutaneous adipocytes of obese individuals, accompanied by a decrease in the mitochondrial DNA transcript COX1, which is associated with reduced mitochondrial function, while the expression of inflammatory pathway genes is upregulated[ 29 ]. The acetyl-CoA derived from BCAA catabolism enters the TCA cycle. The resulting NADH and FADH₂ fuel the electron transport chain, where COX (Complex IV) and other complexes participate in oxidative phosphorylation to generate ATP[ 30 , 31 ]. However, dysfunction of COX1 impedes this entire oxidative phosphorylation process, leading to an accumulation of upstream metabolites, including BCAA breakdown products[ 32 ]. These accumulated metabolites are closely associated with obesity and insulin resistance and, in turn, inhibit further BCAA degradation, thereby promoting the occurrence of CO. In conclusion, our study suggests that these three central genes may play a key role in CO. The transfer of electrons from NADH to ubiquinone, catalyzed by mitochondrial complex I during oxidative phosphorylation, is a crucial step in mitochondrial electron transport. The elevated NADH levels resulting from mitochondrial complex I dysfunction can lead to insufficient ATP synthesis[ 33 ], which may result in increased food intake, exacerbating obesity. An imblance in the NAD+/NADH is also a common occurence in obesity, insulin resistance and diabetes[ 33 ]. To our knowledge, BCAAs catabolism depends on NAD+. The imblance in the NAD+/NADH can inhibit BCAAs catablism, leading to BCAAs accumulation, which further promotes lipogenesis, insulin resistance, and inflammation. The GSEA pathway enrichment analysis showed that the differentially expressed genes were mainly enriched in Mitochondrial electron transport. That suggests they may be important regulatory factors in obesity. Regulatory T cells (Tregs), a specialized subset of CD4 + T cells, are crucial for suppressing immune responses, maintaining tolerance, and preventing autoimmunity[ 34 ]. Recently, it has been highlighted that the number of Tregs in adipose tissue of obese individuals is decreased, and this is attributed to insulin resistance[ 35 , 36 ]. Our immune infiltration analysis showed that PLEK and MIN were most strongly and positively correlated with Tregs. These findings suggest that modulating Tregs may be a potential therapeutic strategy for treating obesity-related metabolic disorders. Although multiple weight-loss medications are currently available, their safety in pediatric populations remains undetermined. Therefore, there is an urgent need to explore potential medications. To screen for potential therapeutic drugs for CO, we conducted molecular regulatory network analysis and drug prediction analysis. MALAT1, one of the long noncoding RNAs (lncRNAs) predicted through molecular regulatory network analysis, is highly conserved. Recent research demonstrated that MALAT1 SNP rs3200401 is associated with the risk of CO in Russian populations[ 37 ]. Moreover, the expression of MALAT1 was positively correlated with BMI and other metabolic syndrome-related parameters, including HOMA-IR, total cholesterol and low-density lipoprotein cholesterol[ 38 ]. Endothelial dysfunction can lead to an increased risk of cadiovascular complication in Obese. Exercise exerted a positive effect against endothelial dysfunction in obese children and adolescents by downregulating MALAT1 expression[ 38 ]. MALAT1 may be a potential therapeutic target. Our drug prediction analysis also showed that 15, 12, and 15 potential targeted drugs were predicted for COX1, NIN, and PLEK, respectively. Quercetin is a kind of natural flavonoid found, has a wide range of physiology effects, such as anti-flammatory, antioxidant, antidiabetic and anticancer functions[ 39 ]. Quercetin is being explored as a treatment for obesity. Studies have revealed that quercetin alleviates obesity by regulating gut microbiota[ 40 , 41 ], which elevates IPA level to activate AhR/IL-22 pathway[ 40 ]. Another study showed that quercetin reverses obesity through upregulating the expression of 12α-hydroxylase (CYP8B1), thereby facilitating cholesterol conversion to cholic acid[ 42 ]. Taken together, quercetin holds a siginificant potential therapeutic value in the treatment of obesity. Further research could explore the mechanisms, providing more effective strategies for the treatment of obesity. COX1 downregulation in CO suggests impaired mitochondrial respiration, reducing ATP production and disrupting energy homeostasis, thereby promoting fat accumulation. Concurrently, elevated PLEK indicates immune activation, with macrophage infiltration and increased secretion of pro-inflammatory cytokines (e.g., TNF-α, IL-6), exacerbating chronic inflammation in adipose tissue. Our study demonstrated that the expression of COX1 is downregulated in CO, suggesting a potentially significant impairment of mitochondrial respiratory function, which in turn reduces ATP production. This reduction can promote energy storage and inhibit catabolic processes, thus exacerbating the progression of obesity. In contrast, PLEK expression is elevated in the CO group, which may suggest the activation and infiltration of immune cells, particularly macrophages, into adipose tissue. M1-type macrophages secrete potent pro-inflammatory cytokines such as TNF-α, IL-6, and MCP-1, which directly intensify the chronic low-grade inflammation in adipose tissue. Based on bioinformatics analysis and machine learning, we systematically identified three related candidate genes (PLEK, MIN and COX1). This study may facilitate the exploration of molecular mechanisms in CO, particularly regarding the immune response and drug action. The screened genes could be used for clinical diagnosis and treatment. However, the use of a bioinformatics approach to explore the molecular mechanism and predict potential therapeutic compounds has limitations, and further lab studies are needed to support these results. 4. Methods 4.1 Data source Two training sets (GSE205668 and GSE87493) related to CO in this study were obtained from the Gene Expression Omnibus (GEO) database ( https://www.ncbi.nlm.nih.gov/geo/ ). The training set GSE205668 (sequencing type: high-throughput; sequencing platform: GPL16791; download time: May 28th, 2025) contained 26 subcutaneous adipose tissue samples from obese children (CO group) and 35 subcutaneous adipose tissue samples from normal controls (control group). The training set GSE87493 (sequencing type: microarray; sequencing platform: GPL6244; download time: May 28th, 2025) contained 12 blood samples from obese children (CO group) and 20 blood samples from normal controls (control group). In addition, 20 branched-chain amino acid metabolism (BCAAM)-related genes (BCAAM-RGs) were obtained from the molecular signatures database (MSigDB) ( https://www.gsea-msigdb.org ) using "REACTOME_BRANCHED_CHAIN_AMINO_ACID_CATABOLISM" as the search keyword[ 43 ]. 4.2 Differential expression analysis To screen out differentially expressed genes (DEGs) in CO, differential expression analysis was first performed on the samples from the CO group and the control group in the training set GSE205668 using the R package "DESeq2" (v 1.48.1)[ 44 ]. The first set of DEGs was identified and named DEGs1. Meanwhile, differential expression analysis was conducted on the samples from the CO group and the control group in the training set GSE87493 using the R package "limma" (v 3.58.1)[ 45 ], and the second set of DEGs was identified and named DEGs2. The screening criteria for both analyses were consistent, with the statistical significance threshold set at P 0.1. Subsequently, the R package "pheatmap" (v 1.0.12)[ 46 ] was used to separately generate heatmaps of the expression levels of DEGs1 and DEGs2 between the CO group and the control group. In the heatmaps, genes were sorted in descending order of |log 2 FC| values, displaying the top 10 upregulated and top 10 downregulated genes with the most significant expression differences. Additionally, the R package "ggplot2" (v 3.5.1)[ 47 ] was employed to draw volcano plots for DEGs1 and DEGs2. Similarly, genes in the volcano plots were sorted in descending order of |log 2 FC| values, and the top 10 upregulated and top 10 downregulated genes with the most significant expression differences were labeled to visually present the overall differential expression patterns. 4.3 Weighted gene co-expression network analysis (WGCNA) To investigate the expression module characteristics of branched-chain amino acids metabolism-related genes (BCAAM-RGs) in CO, first, based on all samples from the training set GSE205668, the BCAAM-RGs score for each sample was calculated using the single-sample gene set enrichment analysis (ssGSEA) algorithm in the R package "GSVA" (v 1.50.5)[ 48 ]. Differences in BCAAM-RGs scores between the CO group and the control group were compared via the Wilcoxon rank-sum test (with a significance threshold of P < 0.05). Subsequently, to identify co-expression module genes closely related to BCAAM, weighted gene co-expression network analysis (WGCNA) was performed based on the BCAAM-RGs scores. Hierarchical clustering analysis was conducted on all samples in the training set GSE205668 using the "hclust" function in the R package "GSVA" (v 1.50.5)[ 48 ]; outlier samples were identified and excluded to obtain a high-quality sample set for constructing the co-expression network, and a sample clustering tree was plotted. Next, the "pickSoftThreshold" function was used to screen the optimal power value within the range of soft threshold (power) parameters 1–20, with the requirement that the scale-free network evaluation coefficient satisfy 0.8 ≤ R² ≤ 0.9. A soft threshold screening plot was drawn to determine the correlation threshold between genes that met the scale-free distribution characteristics. The gene expression adjacency matrix was converted into a topological overlap matrix (TOM), and hierarchical clustering was performed via the "hierarchicalCluster" function to construct a gene clustering tree. Co-expression modules were identified using the dynamic tree cut algorithm. Finally, the correlation between each module and the BCAAM-RGs score was analyzed, with the threshold set as |correlation coefficient (cor)| > 0.3 and P < 0.05. A module-trait correlation heatmap was plotted to visualize the results. One key module significantly positively correlated with the BCAAM-RGs score, and one significantly negatively correlated was selected, and all genes in these two modules were extracted as BCAAM-related key module genes. 4.4 Acquisition of candidate genes and functional enrichment analysis To screen for core DEGs with consistent expression across datasets, the upregulated DEGs (DEGs1-up) from the training set GSE205668 were intersected with the upregulated DEGs (DEGs2-up) from the training set GSE87493. Meanwhile, the downregulated DEGs (DEGs1-down and DEGs2-down) from the two datasets were intersected. Subsequently, the intersected genes from these two parts were merged to obtain DEGs (DEGs3). Thereafter, to identify genes associated with both CO and BCAAM, the R package "ggvenn" (v 0.1.10) ( https://CRAN.R-project.org/package=ggvenn ) was used to intersect DEGs3 with BCAAM-related key module genes, and the resulting candidate genes were obtained for subsequent functional validation. The R package "clusterProfiler" (v 4.10.1)[ 49 ] was employed to perform Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analyses on the candidate genes (adjusted (adj). P < 0.05), aiming to explore their biological functions and mechanisms. GO consisted of 3 parts, namely biological process (BP), cellular component (CC), and molecular function (MF). Additionally, the search tool for recurring instances of neighbouring genes (STRING) ( https://string-db.org/ ) was used to analyze interactions between proteins encoded by candidate genes, with the threshold set as interaction score > 0.4. Finally, the results were visualized by constructing a protein-protein interactions (PPI) network diagram using Cytoscape software (v 3.10.0)[ 50 ]. 4.5 Acquisition of biomarkers To further screen for biomarkers with potential diagnostic value in CO, least absolute shrinkage and selection operator (LASSO) regression analysis, support vector machine-recursive feature elimination (SVM-RFE) analysis, and random forest (RF) analysis were employed to screen biomarkers. First, LASSO regression analysis was performed on the training set GSE205668 using the R package "glmnet" (v 4.1-8)[ 51 ]. The optimal regularization parameter lambda corresponding to the minimum mean squared error was determined through 5-fold cross-validation, and genes with non-zero regression coefficients were screened to obtain feature gene set 1. Meanwhile, regression coefficient plots and cross-validation error curves were generated. Subsequently, SVM-RFE analysis was conducted on the candidate genes using the R package "caret" (v 4.7–1.1)[ 52 ]. With a 5-fold cross-validation and a stepwise decrement strategy (removing 1 feature in each iteration) set, the genes corresponding to the model with the highest cross-validation accuracy were selected as feature gene set 2. Furthermore, RF analysis was performed on the candidate genes using the R package "randomForest" (v 4.7–1.2)[ 52 ]. The parameter was set to a decision tree number of ntree = 100, and the top 10 genes ranked by importance were screened based on the criterion of IncNodePurity > 1 to obtain feature gene set 3. Finally, the intersection of the three feature gene sets was obtained by using the R package "ggvenn" (v 0.1.10) ( https://CRAN.R-project.org/package=ggvenn ), and biomarkers were acquired. 4.6 Construction and evaluation of a nomogram To evaluate the reliability of biomarkers in predicting CO, a nomogram prediction model based on biomarkers was constructed using the R package "rms" (v 8.0–0) ( http://cran.r-project.org/web/packages/rms/ ) with all samples from the training set GSE205668. The scale range of the line corresponding to each biomarker represents its predicted score interval (Points). The total score (Total Points) is obtained by summing the scores of each biomarker, and a higher total score indicates a higher probability of developing CO. The length of the line reflects the contribution weight of the biomarker to the outcome event. Further, the R package "rms" (v 8.0–0) ( http://cran.r-project.org/web/packages/rms/ ) was used to draw a calibration curve to verify the prediction accuracy of the model. The Hosmer-Lemeshow test (HL test) was applied to judge the consistency between the predicted probability and the actual probability. If the P value of the test is greater than 0.05, it indicates that there is no significant difference between the predicted value and the actual value, suggesting that the nomogram model has high calibration accuracy in predicting the incidence of CO. Subsequently, the nomogram's predictive accuracy was assessed via receiver operating characteristic (ROC) curve analysis, performed using the R package "pROC" (v 1.19.0.1) ( https://xrobin.github.io/pROC/ ). The area under the ROC curve (AUC), ranging from 0 to 1, was employed as a metric of discriminatory power, with values exceeding 0.7 considered indicative of good predictive performance. Decision curve analysis (DCA) is a statistical method for evaluating the net benefit of clinical interventions across different probability thresholds, was conducted using the R package "rmda" (v 1.6)[ 53 ]. The DCA curve was created to visualize the nomogram's net benefit across a spectrum of clinically relevant threshold probabilities, with higher nomogram model curve indicating greater clinical utility and improved model performance. 4.7 Correlation analysis and functional similarity analysis of biomarkers To analyze the correlation between biomarkers, correlation analysis of the biomarkers was performed using the R package "psych" (v 2.4.3)[ 54 ], with the threshold for significant correlation set as |cor| > 0.3 and P < 0.05. The results were visualized by drawing a correlation heatmap using the R package "ggplot2" (v 3.5.1)[ 47 ]. To further evaluate the functional similarity among biomarkers, the semantic similarity of the biomarkers in the three main ontological categories of Gene Ontology (GO) — biological process (BP), cellular component (CC), and molecular function (MF) — was calculated using the R package "GOSemSim" (v 2.34.0) ( https://github.com/YuLab-SMU/GOSemSim/issues ). The calculated average semantic similarity score was used to measure the level of similarity, and a score greater than 0.5 was set as the threshold for high functional similarity. 4.8 GeneMANIA analysis and localization analysis of biomarkers To deeply analyze the interaction and functional association between biomarkers and their functionally similar genes, the biomarkers were imported into the GeneMANIA online tool ( http://genemania.org/ ), with the specified analysis species set as "Homo sapiens". Information from multiple biological databases, including that on co-expression, co-localization, protein-protein interactions, and co-participation in pathways, was integrated by GeneMANIA. Finally, a gene interaction network diagram was constructed to visually display the complex associations between biomarkers. Further, the "RCircos" package (v 1.18.4)[ 55 ] was employed to perform chromosomal localization analysis and generate circular genome maps. By integrating the genomic location information of biomarkers, a visualized chromosome map was generated to visually show the distribution of each biomarker on the chromosome. Then, explore the subcellular localization of biomarkers in the cells. First of all, from the National Center for Biotechnology Information (NCBI) database, to obtained the corresponding FASTA sequences based on biomarkers. Then, the FASTA sequences were imported into the mRNALocater database ( http://bio-bigdata.cn/mRNALocater/ ), which predicts the localization probability of RNA molecules in 5 subcellular compartments (nucleus, cytoplasm, ribosome, mitochondria, and exosomes). 4.9 Gene set enrichment analysis (GSEA) of biomarkers To explore the regulatory pathways or biological functions related to the biomarkers, GSEA was conducted for each biomarker based on the CO group samples and control group samples in the training set GSE205668. The Spearman correlation coefficients between other genes and the biomarkers were calculated respectively using the R package "stats" (v 0.1.0)[ 56 ], and the correlation coefficients were sorted in descending order. The gene set "c5.go.bp.v2024.1.Hs.symbols.gmt" from the Molecular Signatures Database (MsigDB) ( https://www.gsea-msigdb.org/gsea/msigdb/ ) was adopted as the reference gene set. Subsequently, GSEA was performed using the R package "clusterProfiler" (v 4.10.0)[ 49 ], and the significance thresholds were set as |normalized enrichment score (NES)| > 1, False Discovery Rate (FDR) < 0.25 and P < 0.05. Finally, significant pathways were selected based on P values, and the top 5 most significant signaling pathways were visualized. 4.10 Immune infiltration analysis CO is closely linked to immune dysregulation, typically manifesting as chronic low-grade inflammation and altered immune cell function[ 57 , 58 ]. To evaluate differences in immune infiltration levels between the CO group and the control group, the relative infiltration abundances of 28 immune cells were calculated using the ssGSEA algorithm[ 59 ] from the R package "GSVA" (v 2.2.0)[ 60 ] based on all samples from the training set GSE205668, and visualized using the R package "ggplot2" (v. 3.5.1)[ 47 ]. Furthermore, differences in the infiltration abundance of each immune cell between the two groups were analyzed using the Wilcoxon rank-sum test. Immune cells meeting the significance criterion (P < 0.05) were defined as differential immune cells, and box plots of their infiltration proportions were plotted and visualized using the R package "ggplot2" (v. 3.5.1)[ 47 ]. Finally, spearman correlation analyses were performed between differential immune cells, and between biomarkers and differential immune cells, respectively, using the R package "psych" (v. 2.4.3)[ 54 ]. The thresholds for correlation significance were set as |cor| > 0.3 and P < 0.05. 4.11 Molecular regulatory network analysis of biomarkers To further explore the upstream regulatory mechanisms of biomarkers, the molecular regulatory network was constructed based on the competing endogenous RNA (ceRNA) theory. Firstly, microRNAs (miRNAs) interacting with biomarkers (mRNAs) were jointly predicted using the diana_microt ( https://dianalab.e-ce.uth.gr/html/dianauniverse/index.php?r=microT_CDS ) and targetscan ( https://www.targetscan.org/ ) databases, and only miRNAs present in the prediction results of both databases were retained as candidate miRNAs. Subsequently, long non-coding RNAs (LncRNAs) interacting with candidate miRNAs were predicted using the StarBase database ( Http://starbase.sysu.edu.cn ). Finally, the LncRNA–miRNA–mRNA regulatory network was constructed using the screened LncRNAs and miRNAs, and biomarkers (mRNAs), and the results were visualized using Cytoscape software (v 3.10.0)[ 50 ]. 4.12 Drug prediction analysis of biomarkers To identify potential drugs for the treatment of CO, drug prediction analyses for each biomarker were performed using the drug signatures database (DsigDB) ( http://dsigdb.tanlab.org/ ). Based on the prediction results, the potential therapeutic drug that might interact with the biomarkers were identified. Finally, a drug-biomarker interaction network was constructed and visualized using Cytoscape software (v 3.10.0)[ 50 ]. 4.13 Reverse transcription quantitative polymerase chain reaction (RT-qPCR) To validate the expression of biomarkers in clinical samples, RT-qPCR analysis was performed. Specifically, a total of 10 blood samples (5 CO and 5 control) were acquired from the clinic in the Fujian Children’s Hospital. Written informed consent was obtained from the parents or legal guardians of all minor participants prior to enrollment in the study. The study had the approval of the Fujian Children’s Hospital ethics committee (approval number: 2025ETKLR10009). Total RNA from the 10 tissue samples was extracted with the TRIzol reagent (Ambion, USA) according to the manufacturer's protocol. Then the RNA concentration was tested using NanoPhotometer N50. The cDNA was synthesized by reverse transcription using the SureScript First-Strand cDNA Synthesis Kit, and the reaction was performed with S1000TM Thermal Cycler (Bio-Rad, USA). The sequences of all primers can be found in Supplementary Table S1 . The qPCR assay was performed with CFX Connect Real-time Quantitative Fluorescence PCR Instrument (Bio-Rad, USA) (pre-denaturation at 95℃ for 1 min, denaturation at 95℃ for 20s, annealing at 55℃ for 20s, extension at 72℃ for 30s, a total of 40 cycles). The relative quantification of mRNAs was calculated using the 2 −ΔΔCT method. The results from the RT-qPCR were exported to Excel, and then imported into Graphpad Prism 5 ( https://www.graphpad.com/ ) for statistical analysis and visualization. 4.14 Statistical analysis The bioinformatics analysis incorporated R software (v 4.4.3) for statistical processing. Statistical significance was defined as a P < 0.05. The Wilcoxon rank sum test was used for between-group comparisons, whereas the t-test was applied to analyze RT-qPCR results across groups. Abbreviations Abbreviation Full term adj. P adjusted P value AhR aryl hydrocarbon receptor ATP adenosine triphosphate AUC area under the curve BAT brown adipose tissue BCAA branched-chain amino acid BCAAM branched-chain amino acid metabolism BCAAM-RG branched-chain amino acid metabolism-related gene BCKA branched-chain keto acid BCAT1 branched-chain amino acid transaminase 1 BCAT2 branched-chain amino acid transaminase 2 BMI body mass index BP biological process CC cellular component cDNA complementary DNA ceRNA competing endogenous RNA CO childhood obesity cor correlation coefficient DCA decision curve analysis DEG differentially expressed gene DsigDB Drug Signatures Database FADR false discovery rate FC fold change FDR false discovery rate GEO Gene Expression Omnibus GeneMANIA Gene Multiple Association Network Integration Algorithm GO Gene Ontology GSVA gene set variation analysis GSEA gene set enrichment analysis HL test Hosmer-Lemeshow test HOMA-IR homeostasis model assessment of insulin resistance IFNGR1 interferon gamma receptor 1 IL-1β interleukin-1 beta IL-6 interleukin-6 IPA indole-3-propionic acid IR insulin resistance IRS-1 insulin receptor substrate-1 JAK1 Janus kinase 1 KEGG Kyoto Encyclopedia of Genes and Genomes LASSO least absolute shrinkage and selection operator limma linear models for microarray data lncRNA long non-coding RNA LPS lipopolysaccharide MAE mean absolute error MAPK mitogen-activated protein kinase MDSC myeloid-derived suppressor cell MF molecular function miRNA microRNA mTOR mammalian target of rapamycin mTORC1 mammalian target of rapamycin complex 1 MSigDB Molecular Signatures Database NADH nicotinamide adenine dinucleotide, reduced form NES normalized enrichment score PPI protein-protein interaction PPARγ peroxisome proliferator-activated receptor gamma PRDM16 PR domain-containing protein 16 qPCR quantitative polymerase chain reaction RF random forest ROC receiver operating characteristic RT-qPCR reverse transcription quantitative polymerase chain reaction SNP single nucleotide polymorphism ssGSEA single-sample gene set enrichment analysis STAT1 signal transducer and activator of transcription 1 STRING Search Tool for the Retrieval of Interacting Genes/Proteins SVM-RFE support vector machine-recursive feature elimination T2DM type 2 diabetes mellitus TCA tricarboxylic acid cycle TNF-α tumor necrosis factor-alpha TOM topological overlap matrix Treg regulatory T cell WAT white adipose tissue WGCNA weighted gene co-expression network analysis Declarations Acknowledgements We would like to express our sincere gratitude to all individuals and organizations who supported and assisted us throughout this research. Special thanks to my family for their unwavering support and encouragement throughout this work. In conclusion, we extend our thanks to everyone who has supported and assisted us along the way. Without your support, this research would not have been possible. Author contributions Huijie Zhang: Conceptualization, Data curation, Validation, Visualization, Writing–original draft, Writing–review & editing. Ying Lei: Data curation, Validation, Visualization, Writing–review & editing. Limei Guan: Conceptualization, Supervision, Writing–review & editing. Hui Liu: Conceptualization, Project administration, Supervision, Writing–review & editing. This work currently described has not been published, is not being considered for publication elsewhere, and its publication was approved by all authors. Data Availability statement The datasets analyzed in this study can be found in the [Gene Expression Omnibus (GEO) database] [http://www.ncbi.nlm.nih.gov/geo/] under accession numbers GSE205668 and GSE87493. The branched-chain amino acid metabolism-related gene set used in this study was obtained from the [Molecular Signatures Database (MSigDB)] [https://www.gsea-msigdb.org/gsea/msigdb/]. Ethics approval This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Fujian Children’s Hospital ethics committee, and the approval number and date of approval are [2025ETKLR10009] and [2025.02.25]. Informed consent was obtained from the parents or legal guardians of all minor participants included in the study. Consent to publish Not applicable. Conflict of Interest The authors declare no competing interests. Funding The research reported in this project was generously supported by [Startup Fund for scientific research, Fujian Medical University] under grant agreement number [2021QH1196]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. References Pappachan, J. M., Fernandez, C. J. & Ashraf, A. P. Rising tide: The global surge of type 2 diabetes in children and adolescents demands action now. World J Diabetes . 15 , 797–809 (2024). Di Cesare, M. et al. The epidemiological burden of obesity in childhood: a worldwide epidemic requiring urgent action. BMC Med . 17 , 212 (2019). De Spiegeleer, M. et al. Paediatric obesity: a systematic review and pathway mapping of metabolic alterations underlying early disease processes. Mol Med . 27 , 145 (2021). Newgard, C. B. et al. A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance. Cell Metab . 9 , 311–326 (2009). Huang, H., Chen, H., Yao, Y. & Lou, X. Branched-chain amino acids supplementation induces insulin resistance and pro-inflammatory macrophage polarization via INFGR1/JAK1/STAT1 signal pathway. Mol Med . 30 , 149 (2024). Li, H. & Seugnet, L. Decoding the nexus: branched-chain amino acids and their connection with sleep, circadian rhythms, and cardiometabolic health. Neural Regen Res . 20 , 1350–1363 (2025). Ma, Q. X. et al. BCAA-BCKA axis regulates WAT browning through acetylation of PRDM16. Nat Metab . 4 , 106–122 (2022). Gumus Balikcioglu, P. et al. Branched-chain α-keto acids and glutamate/glutamine: Biomarkers of insulin resistance in childhood obesity. Endocrinol Diabetes Metab . 6 , e388 (2023). Verkerke, A. R. P. et al. BCAA-nitrogen flux in brown fat controls metabolic health independent of thermogenesis. Cell . 187 , 2359–2374.e2318 (2024). Ma, A. D., Brass, L. F. & Abrams, C. S. Pleckstrin associates with plasma membranes and induces the formation of membrane projections: requirements for phosphorylation and the NH2-terminal PH domain. J Cell Biol . 136 , 1071–1079 (1997). Alim, M. A. et al. Pleckstrin Levels Are Increased in Patients with Chronic Periodontitis and Regulated via the MAP Kinase-p38α Signaling Pathway in Gingival Fibroblasts. Front Immunol . 12 , 801096 (2021). Xu, Z., Wen, C. & Wang, W. Role of MAPK and PI3K-Akt signaling pathways in cuprizone-induced demyelination and cognitive impairment in mice. Behav Brain Res . 458 , 114755 (2024). Huang, X., Liu, G., Guo, J. & Su, Z. The PI3K/AKT pathway in obesity and type 2 diabetes. Int J Biol Sci . 14 , 1483–1496 (2018). Wang, K. et al. The Effect of Enteric-Derived Lipopolysaccharides on Obesity. Int J Mol Sci . 25 (2024). Song, B. et al. Association of the gut microbiome with fecal short-chain fatty acids, lipopolysaccharides, and obesity in young Chinese college students. Front Nutr . 10 , 1057759 (2023). Hertiš Petek, T., Homšak, E., Svetej, M. & Marčun Varda, N. Systemic Inflammation and Oxidative Stress in Childhood Obesity: Sex Differences in Adiposity Indices and Cardiovascular Risk. Biomedicines . 13 (2024). González-Domínguez, Á. et al. Altered Metal Homeostasis Associates with Inflammation, Oxidative Stress, Impaired Glucose Metabolism, and Dyslipidemia in the Crosstalk between Childhood Obesity and Insulin Resistance. Antioxidants (Basel) . 11 (2022). Estébanez, B., Huang, C. J., Rivera-Viloria, M., González-Gallego, J. & Cuevas, M. J. Exercise Outcomes in Childhood Obesity-Related Inflammation and Oxidative Status. Front Nutr . 9 , 886291 (2022). Martínez de Morentin, P. B. et al. Hypothalamic mTOR: the rookie energy sensor. Curr Mol Med . 14 , 3–21 (2014). Zhang, W. et al. Modulation of food intake by mTOR signalling in the dorsal motor nucleus of the vagus in male rats: focus on ghrelin and nesfatin-1. Exp Physiol . 98 , 1696–1704 (2013). Dong, P. et al. Dampened VEPH1 activates mTORC1 signaling by weakening the TSC1/TSC2 association in hepatocellular carcinoma. J Hepatol . 73 , 1446–1459 (2020). Kimball, S. R., Gordon, B. S., Moyer, J. E., Dennis, M. D. & Jefferson, L. S. Leucine induced dephosphorylation of Sestrin2 promotes mTORC1 activation. Cell Signal . 28 , 896–906 (2016). Wu, X. et al. FLCN Maintains the Leucine Level in Lysosome to Stimulate mTORC1. PLoS One . 11 , e0157100 (2016). Chen, J. et al. SAR1B senses leucine levels to regulate mTORC1 signalling. Nature . 596 , 281–284 (2021). Cangelosi, A. L. et al. Zonated leucine sensing by Sestrin-mTORC1 in the liver controls the response to dietary leucine. Science . 377 , 47–56 (2022). Wu, D., Rawal, K., Eeda, V., Lim, H. Y. & Wang, W. Identification and Sorting of Adipose Inflammatory and Metabolically Activated Macrophages in Diet-Induced Obesity. Bio Protoc . 15 , e5479 (2025). Omer, S., Li, J., Yang, C. X. & Harrison, R. E. Ninein promotes F-actin cup formation and inward phagosome movement during phagocytosis in macrophages. Mol Biol Cell . 35 , ar26 (2024). Papathanassiu, A. E. et al. BCAT1 controls metabolic reprogramming in activated human macrophages and is associated with inflammatory diseases. Nat Commun . 8 , 16040 (2017). Heinonen, S. et al. Mitochondria-related transcriptional signature is downregulated in adipocytes in obesity: a study of young healthy MZ twins. Diabetologia . 60 , 169–181 (2017). Gladyck, S., Aras, S., Hüttemann, M. & Grossman, L. I. Regulation of COX Assembly and Function by Twin CX(9)C Proteins-Implications for Human Disease. Cells . 10 (2021). Wu, G. J. et al. Genistein Triggers Translocation of Estrogen Receptor-Alpha in Mitochondria to Induce Expressions of ATP Synthesis-Associated Genes and Improves Energy Production and Osteoblast Maturation. Am J Chin Med . 49 , 901–923 (2021). Adlimoghaddam, A. & Albensi, B. C. The nuclear factor kappa B (NF-κB) signaling pathway is involved in ammonia-induced mitochondrial dysfunction. Mitochondrion . 57 , 63–75 (2021). Bae, H. R. et al. D-Allulose Ameliorates Dysregulated Macrophage Function and Mitochondrial NADH Homeostasis, Mitigating Obesity-Induced Insulin Resistance. Nutrients . 15 (2023). Chen, X. et al. Differential roles of human CD4(+) and CD8(+) regulatory T cells in controlling self-reactive immune responses. Nat Immunol . 26 , 230–239 (2025). Bradley, D. et al. Interferon gamma mediates the reduction of adipose tissue regulatory T cells in human obesity. Nat Commun . 13 , 5606 (2022). Soedono, S. et al. Obese visceral adipose dendritic cells downregulate regulatory T cell development through IL-33. Front Immunol . 15 , 1335651 (2024). Shkurat, T. P. et al. The Role of Genetic Variants in the Long Non-Coding RNA Genes MALAT1 and H19 in the Pathogenesis of Childhood Obesity. Noncoding RNA . 9 (2023). Zhao, W., Yin, Y., Cao, H. & Wang, Y. Exercise Improves Endothelial Function via the lncRNA MALAT1/miR-320a Axis in Obese Children and Adolescents. Cardiol Res Pract . 2021 , 8840698 (2021). Shams, E. et al. The effect of quercetin on obesity and reproduction through the expression of genes involved in the hypothalamus-pituitary-gonadal axis. JBRA Assist Reprod . 29 , 211–218 (2025). Lu, J. et al. Quercetin ameliorates obesity and inflammation via microbial metabolite indole-3-propionic acid in high fat diet-induced obese mice. Front Nutr . 12 , 1574792 (2025). Zhao, L. et al. Quercetin Ameliorates Gut Microbiota Dysbiosis That Drives Hypothalamic Damage and Hepatic Lipogenesis in Monosodium Glutamate-Induced Abdominal Obesity. Front Nutr . 8 , 671353 (2021). Liu, J. et al. Quercetin-Driven Akkermansia Muciniphila Alleviates Obesity by Modulating Bile Acid Metabolism via an ILA/m(6)A/CYP8B1 Signaling. Adv Sci (Weinh) . 12 , e2412865 (2025). Neinast, M., Murashige, D. & Arany, Z. Branched Chain Amino Acids. Annu Rev Physiol . 81 , 139–164 (2019). Li, X., Fan, K. & Yang, B. Dynamic transcriptomic landscape of myogenesis in Muscovy ducks (Cairina moschata): integrative analysis of hub genes post-hatching. Anim Biosci . 39 , 250159 (2026). Ritchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res . 43 , e47 (2015). Mohammed, R., Nader, S. M., Hamza, D. A. & Sabry, M. A. Public health implications of multidrugresistant and methicillinresistant Staphylococcus aureus in retail oysters. Sci Rep . 15 , 4496 (2025). Purssell, E., Frood, S. & Sagoo, R. Beyond the labels: Classifying countries by child health outcomes - A cluster analysis of child mortality and child-health data. Glob Health Action . 18 , 2526315 (2025). Yang, H. et al. Molecular targets of neuroplasticity in ischemic stroke: insights from GEO database, single-cell analysis and immune infiltration analysis. Front Aging Neurosci . 17 , 1561282 (2025). Wu, T. et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation (Camb) . 2 , 100141 (2021). Otasek, D., Morris, J. H., Bouças, J., Pico, A. R. & Demchak, B. Cytoscape Automation: empowering workflow-based network analysis. Genome Biol . 20 , 185 (2019). Friedman, J., Hastie, T. & Tibshirani, R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw . 33 , 1–22 (2010). Mishra, A., Maiti, R., Jena, M. & Srinivasan, A. Evaluating machine learning algorithms for prediction of treatment response for sleep disturbances in patients with schizophrenia: A post-hoc analysis from a randomized controlled trial. Psychiatr Danub . 37 , 46–54 (2025). Hull, T. D. et al. Cyclic Di-GMP phosphodiesterases RmdA and RmdB are involved in regulating colony morphology and development in Streptomyces coelicolor. J Bacteriol . 194 , 4642–4651 (2012). Kasyanov, E. D., Yakovleva, Y. V., Mudrakova, T. A., Kasyanova, A. A. & Mazo, G. E. [Comorbidity patterns and structure of depressive episodes in patients with bipolar disorder and major depressive disorder]. Zh Nevrol Psikhiatr Im S S Korsakova . 123 , 108–114 (2023). Yang, Z. et al. Dicoumarol sensitizes hepatocellular carcinoma cells to ferroptosis induced by imidazole ketone erastin. Front Immunol . 16 , 1531874 (2025). Yao, Z. et al. Deciphering the multidimensional impact of IGFBP1 expression on cancer prognosis, genetic alterations, and cellular functionality: A comprehensive Pan-cancer analysis. Heliyon . 10 , e37402 (2024). Dai, W. et al. Influence of adipose tissue immune dysfunction on childhood obesity. Cytokine Growth Factor Rev . 65 , 27–38 (2022). Lagou, M. K. & Karagiannis, G. S. Obesity-induced thymic involution and cancer risk. Semin Cancer Biol . 93 , 3–19 (2023). Qiu, C. et al. Identification and verification of XDH genes in ROS induced oxidative stress response of osteoarthritis based on bioinformatics analysis. Sci Rep . 15 , 29759 (2025). Cao, L. et al. Integrative Analysis of Novel Ferroptosis-Related Genes Signatures as Prognostic Biomarkers in Ovarian Cancer. Cancer Rep (Hoboken) . 8 , e70284 (2025). Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviews received at journal 03 May, 2026 Reviewers agreed at journal 26 Apr, 2026 Reviewers agreed at journal 26 Apr, 2026 Reviewers invited by journal 24 Apr, 2026 Editor assigned by journal 24 Apr, 2026 Editor invited by journal 17 Apr, 2026 Submission checks completed at journal 15 Apr, 2026 First submitted to journal 15 Apr, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9238869","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":634316796,"identity":"f361a481-6d24-481c-9b88-54b22abbfb7f","order_by":0,"name":"Huijie Zhang","email":"","orcid":"","institution":"Fujian Children's Hospital (Fujian Branch of Shanghai Children's Medical Center), College of Clinical Medicine for Obstetrics \u0026 Gynecology and Pediatrics, Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Huijie","middleName":"","lastName":"Zhang","suffix":""},{"id":634316797,"identity":"361ccf24-6ebf-4f3a-9ada-a1ffed13a7d8","order_by":1,"name":"Ying Lei","email":"","orcid":"","institution":"Fujian Children's Hospital (Fujian Branch of Shanghai Children's Medical Center), College of Clinical Medicine for Obstetrics \u0026 Gynecology and Pediatrics, Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Ying","middleName":"","lastName":"Lei","suffix":""},{"id":634316798,"identity":"ce77de3e-8977-4ca9-b4e1-5667df301567","order_by":2,"name":"Limei Guan","email":"","orcid":"","institution":"Fujian Children's Hospital (Fujian Branch of Shanghai Children's Medical Center), College of Clinical Medicine for Obstetrics \u0026 Gynecology and Pediatrics, Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Limei","middleName":"","lastName":"Guan","suffix":""},{"id":634316799,"identity":"e8c17388-721c-4672-9dd0-33ca42b06b46","order_by":3,"name":"Hui Liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAElEQVRIiWNgGAWjYDACCSBmbGDgYWxmYHxQwUOiFmaDM0AthHVBtYAAm8QZBiK0yM9ufvbw647DMsztzMcqDsjUydlL5Bgw/KjYhlML45xj5sayZw4DHcaWduMAz2FjHqAWxp4zt3FqYZZIMJOWbANp4TG7/YHnQGIPUAszYxtuLWwS6d+gWvi/FRzgqasnqAXoDDPJjxBb2BgO8DAn8BDSIiGRUybN2JYO8ouxBNAvhj1nnhUcxOcX+Rnp2yR/tlnbG/YffvjhYE+dPHt78sYHPypwawEHASgqDBuABGMPkBDIMDiAVz1I4Q+QdWAmiMV//AEhHaNgFIyCUTCyAACLb1QuGBwtTAAAAABJRU5ErkJggg==","orcid":"","institution":"Fujian Children's Hospital (Fujian Branch of Shanghai Children's Medical Center), College of Clinical Medicine for Obstetrics \u0026 Gynecology and Pediatrics, Fujian Medical University","correspondingAuthor":true,"prefix":"","firstName":"Hui","middleName":"","lastName":"Liu","suffix":""}],"badges":[],"createdAt":"2026-03-27 01:38:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9238869/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9238869/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108533723,"identity":"f02e77ad-9ea6-4394-a5fe-6d05c94d387e","added_by":"auto","created_at":"2026-05-05 16:36:11","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":235937,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification and functional annotation of candidate genes related to branched-chain amino acid metabolism in childhood obesity.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e Heatmap showing the top 10 upregulated and top 10 downregulated differentially expressed genes (DEGs) in GSE205668. \u003cstrong\u003eb\u003c/strong\u003e Volcano plot of DEGs identified in GSE205668. \u003cstrong\u003ec\u003c/strong\u003e Heatmap showing the top 10 upregulated and top 10 downregulated DEGs in GSE87493. \u003cstrong\u003ed\u003c/strong\u003eVolcano plot of DEGs identified in GSE87493. \u003cstrong\u003ee\u003c/strong\u003e Venn diagram of co-upregulated genes shared by GSE205668 and GSE87493. \u003cstrong\u003ef\u003c/strong\u003e Venn diagram of co-downregulated genes shared by GSE205668 and GSE87493. \u003cstrong\u003eg\u003c/strong\u003eComparison of branched-chain amino acid metabolism-related gene (BCAAM-RG) scores between the control and childhood obesity (CO) groups. \u003cstrong\u003eh\u003c/strong\u003eHierarchical clustering dendrogram of samples in GSE205668. \u003cstrong\u003ei\u003c/strong\u003eAnalysis of soft-thresholding power for weighted gene co-expression network analysis (WGCNA). \u003cstrong\u003ej\u003c/strong\u003e Gene dendrogram and module assignment identified by dynamic tree cutting. \u003cstrong\u003ek\u003c/strong\u003e Module-trait relationship heatmap showing the correlations between co-expression modules and BCAAM-RG scores. \u003cstrong\u003el\u003c/strong\u003e Venn diagram showing the intersection between DEGs3 and BCAAM-related key module genes. \u003cstrong\u003em\u003c/strong\u003e Gene Ontology (GO) enrichment analysis of the 34 candidate genes. \u003cstrong\u003en\u003c/strong\u003eKyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of the 34 candidate genes. \u003cstrong\u003eo\u003c/strong\u003e Protein-protein interaction (PPI) network of the 34 candidate genes constructed using the STRING database and visualized in Cytoscape.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-9238869/v1/783d23522bd7b098062e3489.png"},{"id":108804998,"identity":"cc528bff-c1c7-4381-b8c6-ea60f88a34a6","added_by":"auto","created_at":"2026-05-08 15:24:28","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":167428,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMachine learning-based identification of biomarkers and construction of a nomogram for childhood obesity.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e Least absolute shrinkage and selection operator (LASSO) coefficient profiles of candidate genes. \u003cstrong\u003eb\u003c/strong\u003eFive-fold cross-validation for tuning parameter selection in the LASSO model. \u003cstrong\u003ec\u003c/strong\u003eSupport vector machine-recursive feature elimination (SVM-RFE) analysis showing the optimal number of features. \u003cstrong\u003ed\u003c/strong\u003e Distribution of gene importance scores in the random forest (RF) model. \u003cstrong\u003ee\u003c/strong\u003e Top 10 important genes identified by RF analysis. \u003cstrong\u003ef\u003c/strong\u003e Venn diagram showing the overlap among genes selected by LASSO, SVM-RFE, and RF analyses. \u003cstrong\u003eg\u003c/strong\u003eNomogram constructed based on PLEK, NIN, and COX1 for predicting childhood obesity. \u003cstrong\u003eh\u003c/strong\u003e Calibration curve of the nomogram model. \u003cstrong\u003ei\u003c/strong\u003eReceiver operating characteristic (ROC) curve of the nomogram model. \u003cstrong\u003ej\u003c/strong\u003eDecision curve analysis (DCA) of the nomogram model.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-9238869/v1/07014152a58c25b05ef7770d.png"},{"id":108533725,"identity":"83d8df9d-4f73-4115-a959-6d8a12a56298","added_by":"auto","created_at":"2026-05-05 16:36:11","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":340356,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCorrelation, functional similarity, and GeneMANIA analyses of the identified biomarkers.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e Correlation heatmap among PLEK, NIN, and COX1. \u003cstrong\u003eb\u003c/strong\u003eFunctional similarity analysis of the three biomarkers based on GO semantic similarity. \u003cstrong\u003ec\u003c/strong\u003e GeneMANIA interaction network and enriched functional categories of the biomarkers and their functionally related genes.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-9238869/v1/2a8fec80f119a6fedf3a19aa.png"},{"id":108533726,"identity":"16004d74-839a-4495-af13-7f27e856064f","added_by":"auto","created_at":"2026-05-05 16:36:12","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":196916,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLocalization analysis and gene set enrichment analysis (GSEA) of the identified biomarkers.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e Chromosomal localization of PLEK and NIN. \u003cstrong\u003eb\u003c/strong\u003ePredicted subcellular localization of PLEK, NIN, and COX1. \u003cstrong\u003ec\u003c/strong\u003eGSEA plot of PLEK. \u003cstrong\u003ed\u003c/strong\u003e GSEA plot of NIN. \u003cstrong\u003ee\u003c/strong\u003e GSEA plot of COX1.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-9238869/v1/6331306fec7d86d651b6b2c1.png"},{"id":108803977,"identity":"023e5e04-086f-4005-be87-5d221a2a09d0","added_by":"auto","created_at":"2026-05-08 15:13:18","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":310005,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eImmune infiltration characteristics and their correlations with biomarkers in childhood obesity.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e Landscape of infiltration abundances of 28 immune cell types in the control and CO groups. \u003cstrong\u003eb\u003c/strong\u003eDifferentially infiltrated immune cells between the control and CO groups. \u003cstrong\u003ec\u003c/strong\u003eCorrelation heatmap among differential immune cells. \u003cstrong\u003ed\u003c/strong\u003eCorrelation heatmap between the three biomarkers and differential immune cells.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-9238869/v1/fca37355d5257256a0edbb3a.png"},{"id":108804776,"identity":"c01121fa-bf9c-4532-bc16-5449bb609999","added_by":"auto","created_at":"2026-05-08 15:23:28","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":211678,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMolecular regulatory network and drug prediction analyses of the identified biomarkers.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e Competing endogenous RNA (ceRNA) regulatory network composed of lncRNAs, miRNAs, and mRNAs centered on the identified biomarkers. \u003cstrong\u003eb\u003c/strong\u003eDrug-gene interaction network showing the predicted candidate drugs targeting PLEK, NIN, and COX1.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-9238869/v1/a2e2069d28e0e5fbb870c702.png"},{"id":108533729,"identity":"33187c60-875a-45e7-85aa-7c9d1c226570","added_by":"auto","created_at":"2026-05-05 16:36:12","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":91766,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eExperimental validation of biomarker expression by RT-qPCR.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003eRelative expression of PLEK in the control and CO groups. \u003cstrong\u003eb\u003c/strong\u003eRelative expression of NIN in the control and CO groups. \u003cstrong\u003ec\u003c/strong\u003eRelative expression of COX1 in the control and CO groups. Data are presented as mean ± standard deviation. *P \u0026lt; 0.05; ns, not significant.\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-9238869/v1/42164dd635135a2765be4d9e.png"},{"id":108809872,"identity":"65c1c16f-33dc-47be-b0d4-7f10833f6aa5","added_by":"auto","created_at":"2026-05-08 15:56:00","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1894851,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9238869/v1/96728699-3bce-4a12-b4cf-44ce4905b8f9.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Exploring biomarkers related to branched-chain amino acid metabolism in childhood obesity based on transcriptomics and experimental verification","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eObesity, characterized by abnormal or excessive fat accumulation, is one of the most severe public health problems of this century. Of particular concern is the increasing prevalence in children and adolescents, with global childhood obesity (CO) figures projected to double by 2035 from 2020 levels to 208\u0026nbsp;million boys (a 100% rise) and 175\u0026nbsp;million girls (a 125% increase)[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Notably, CO is not only recognized as an independent risk factor for adult obesity but also leads to numerous metabolic or nonmetabolic complications, such as early puberty, type 2 diabetes mellitus (T2DM), non-alcoholic fatty liver disease, cardiovascular diseases, asthma, and even cancer[\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Therefore, enhancing early screening and management of CO and thereby enabling timely intervention and treatment through the discovery and validation of novel biomarkers holds significant public health importance for alleviating the future burden of chronic diseases in adulthood.\u003c/p\u003e \u003cp\u003eBranched-Chain Amino Acids (BCAAs; valine, leucine, and isoleucine) are essential amino acids that play a pivotal role in lipid, carbohydrate, and protein metabolism. Studies have highlighted a more complex relationship between BCAAs metabolism and the pathophysiology of metabolic diseases, such as obesity, insulin resistance (IR), and type 2 diabetes. Plasma elevations of BCAAs are widely observed in CO, especially when IR is present[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. MS-based metabolomics identifies elevated BCAAs as key features of obesity-induced IR, correlating more strongly with IR than lipid metabolites[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Mechanistically, BCAAs drive IR by persistent mTORC1 activation and subsequent IRS-1 phosphorylation, a process reversible by rapamycin in rodent models[\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. In a mouse model, BCAA supplementation induced obesity and IR, and promoted adipose tissue macrophage M1 polarization via the IFNGR1/JAK1/STAT1 signaling pathway[\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. The sleep-wake cycle is a critical process for sustaining brain function and overall health, encompassing metabolic and cardiovascular systems. Furthermore, plasma concentrations of BCAAs fluctuate throughout the circadian day. This correlation may be associated with sleep disorders and obesity, and research indicates that BCAAs could be orchestrators between the sleep-wake cycle and metabolism[\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eRecent research has identified a key function of the BCAA-BCKA axis in white adipose tissue (WAT) browning. Acetyl-CoA derived from branched-chain keto acids (BCKAs) suppresses WAT browning by acetylating PR domain-containing protein 16 (PRDM16) and disrupting its interaction with peroxisome proliferator-activated receptor-γ (PPARγ)[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Furthermore, telmisartan promotes WAT browning and alleviates obesity via \u003cem\u003eBCAT2\u003c/em\u003e suppression[\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Research also demonstrated that BCKAs and glutamate/glutamine may be biomarkers of IR in CO[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. However, the exact role of BCAAs catabolism in the pathogenesis of obesity is elusive, coupled with a lack of systematic screening and in-depth mechanistic validation of key regulatory genes in pediatric populations.\u003c/p\u003e \u003cp\u003eTo explore the effect of BCAAs catabolism in CO, we employed a comprehensive strategy by integrating transcriptomics and machine learning to pinpoint key biomarkers, followed by their multi-faceted characterization and experimental validation to assess their diagnostic and therapeutic potential.\u003c/p\u003e"},{"header":"2. Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Acquisition and functional enrichment analysis of 34 candidate genes\u003c/h2\u003e \u003cp\u003eTo identify genes associated with CO, differential expression analysis was first performed between the CO group and the control group in the training set GSE205668, and 4110 differentially expressed genes (DEGs1) were identified, among which 1795 genes were upregulated and 2315 genes were downregulated in the CO group (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea-b, \u003cb\u003eSupplementary Table S2\u003c/b\u003e). Subsequently, differential expression analysis was conducted between the CO group and the control group in the training set GSE87493, and 2149 differentially expressed genes (DEGs2) were obtained, with 838 genes upregulated and 1311 genes downregulated in the CO group (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec-d, \u003cb\u003eSupplementary Table S3\u003c/b\u003e). Genes with consistent expression directions (concomitant upregulation or concomitant downregulation) in both datasets were intersected, ultimately yielding 60 co-upregulated genes and 88 co-downregulated genes, totaling 148 intersecting differentially expressed genes (DEGs3) (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee-f). Subsequently, BCAAM-RGs scores were calculated based on all samples from the training set GSE205668, and results of the Wilcoxon rank-sum test showed that there was a significant difference in BCAAM-RGs scores between the control group and the CO group (P\u0026thinsp;=\u0026thinsp;0.0026) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eg). No obvious outlier samples were identified by hierarchical clustering analysis of samples (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eh\u003cb\u003e)\u003c/b\u003e. Further, WGCNA was applied, and a soft threshold (power) of 12 was selected to construct a scale-free network; at this point, the model goodness of fit R\u0026sup2;\u0026asymp;0.8 and the mean value of the adjacency function approached 0, indicating that the network conformed to the scale-free topological characteristics (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ei\u003cb\u003e)\u003c/b\u003e. Hierarchical clustering and dynamic tree cutting were performed based on the TOM, with the minimum number of module genes set to 100, and 9 modules were finally obtained (excluding the gray module) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ej\u003cb\u003e)\u003c/b\u003e. Correlations between each module and BCAAM-RGs scores were analyzed, and 6 modules were found to be significantly correlated with BCAAM-RGs scores (|cor|\u0026gt;0.3, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Among them, the brown module showed the strongest negative correlation with BCAAM-RGs scores (cor=-0.83, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and the red module showed the strongest positive correlation (cor\u0026thinsp;=\u0026thinsp;0.77, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). After genes were extracted from these two modules, a total of 998 BCAAM-related key module genes were obtained (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ek, \u003cb\u003eSupplementary Tables S4-5\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eAfter taking the intersection of DEGs3 and BCAAM-related key module genes, a total of 34 candidate genes were obtained (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003el, \u003cb\u003eSupplementary Table S6\u003c/b\u003e). To further explore the biological functions and mechanisms of the candidate genes, GO and KEGG enrichment analyses were conducted on the 34 candidate genes, resulting in 127 GO terms (adj.P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003em, \u003cb\u003eSupplementary Table S7\u003c/b\u003e), including 93 biological process (BP) terms such as positive regulation of platelet activation and positive regulation of integrin activation; 23 cellular component (CC) terms such as centriolar subdistal appendage and mitochondrial respiratory chain complex III; and 11 molecular function (MF) terms such as microtubule minus-end binding and cytochrome-c oxidase activity. The KEGG enrichment analysis showed that the candidate genes were mainly involved in 12 signaling pathways (adj.P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), including cardiac muscle contraction, oxidative phosphorylation, and non-alcoholic fatty liver disease, etc (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003en, \u003cb\u003eSupplementary Table S8\u003c/b\u003e). Subsequently, analysis of the interaction relationships between proteins encoded by biomarkers was performed via the STRING database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.string-db.org\u003c/span\u003e\u003cspan address=\"https://www.string-db.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). During data processing, some gene names were automatically converted by the STRING database to internationally accepted mitochondrial gene nomenclature: COX2 was converted to MT-CO2, COX1 to MT-CO1, ND1, ND2, ND3, ND4, and ND5 to MT-ND1, MT-ND2, MT-ND3, MT-ND4, and MT-ND5, respectively, and CYTB to MT-CYB. Results of the analysis showed that close associations existed between mitochondrial respiratory chain components (including MT-ND1, MT-ND2, MT-ND3, MT-ND4, MT-ND5, MT-CO1, MT-CO2, MT-CYB) and PTGS1, suggesting that these proteins may play synergistic roles in energy metabolism processes (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eo\u003cb\u003e)\u003c/b\u003e. The candidate genes and enriched pathways identified in this part can provide a theoretical basis for potential therapeutic targets and mechanism studies of CO.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Identification of 3 biomarkers and construction of a nomogram\u003c/h2\u003e \u003cp\u003eTo further screen reliable biomarkers from 34 candidate genes, LASSO regression analysis, SVM-RFE analysis, and RF analysis were applied in this study. Firstly, LASSO regression analysis identified a feature gene set 1 containing 7 genes at log(lambda.min) = -2.4879, which included EHD4, PLEK, NIN, MAP1B, MYH9, LTB, and COX1 (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea-b\u003cb\u003e)\u003c/b\u003e. Results of SVM-RFE analysis showed that the cross-validation accuracy reached the highest when the number of features was set to 30, and thus a feature gene set 2 containing 30 genes was obtained, including PLEK, PTPN6, MYO9B, ARPC1B, COX1, ALCAM, CYTH4, NIN, etc (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec\u003cb\u003e)\u003c/b\u003e. RF analysis, based on gene importance ranking, selected the top 10 genes in terms of importance from the 34 candidate genes to form feature gene set 3, which included FADD, CTH, ALCAM, NIN, PLEK, PTPN6, ND5, COX1, TBC1D10C, and DOCK2 (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed-e\u003cb\u003e)\u003c/b\u003e. By taking the intersection of the three feature gene sets, 3 biomarkers were obtained, namely PLEK, NIN, and COX1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ef\u003cb\u003e)\u003c/b\u003e. Subsequently, the nomogram model constructed based on the 3 biomarkers showed good predictive performance. For example, when the total points reached 2.08, the predicted probability of CO was 83.2% (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eg). Meanwhile, the performance of the nomogram model was validated using multiple indicators. Results of calibration curve analysis showed that the P value of the Hosmer-Lemeshow (HL) test was 0.85, indicating no significant difference between the predicted values of the model and the actual observed values, with good calibration of the model. In addition, the mean absolute error (MAE) was 0.072, which further confirmed that the error between the actual disease risk and the risk predicted by the model was very small, indicating that the nomogram model had high accuracy in predicting CO (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eh). Results of ROC analysis showed that the AUC value was 0.846, demonstrating that the nomogram model had good discriminative ability in distinguishing CO patients from non-CO patients (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ei). Furthermore, decision curve analysis (DCA) was used to evaluate the clinical utility of the model, which quantified the net benefit of intervention using the prediction model under different risk probability thresholds. The results indicated that when the high-risk threshold was in the range of 0.1\u0026ndash;0.7, the net benefit of the nomogram model curve was significantly higher than that of the negative prediction decision curve (None curve) and the positive prediction decision curve (ALL curve), suggesting that the nomogram model had better predictive performance and application value in clinical practice (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ej). This part of the study systematically screened out biomarkers with potential diagnostic value and constructed a reliable nomogram prediction model, providing a theoretical basis for the early, accurate diagnosis of CO and personalized diagnosis and treatment strategies.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Correlation analysis, functional similarity analysis, and GeneMANIA analysis of biomarkers\u003c/h2\u003e \u003cp\u003eCorrelation analysis of the three biomarkers showed that there were significant correlations between them (|cor| \u0026gt; 0.3, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Among them, the highest positive correlation was observed between NIN and PLEK (cor\u0026thinsp;=\u0026thinsp;0.65, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), and the highest negative correlation was observed between PLEK and COX1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD) (cor=-0.73, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), suggesting that they may exert a synergistic regulatory effect in the occurrence and development of CO (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea). Results of functional similarity analysis indicated that the average semantic similarity scores of COX1, PLEK, and NIN were all less than 0.5, indicating low functional similarity among the three. Among them, the scores of PLEK and NIN were relatively close (both approximately 0.400), indicating weak functional similarity between the two (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb). To further explore the functional associations of these biomarkers, the GeneMANIA tool was used in this study to analyze their potential interaction network with functionally similar genes (COX1 had been converted to its official gene symbol MT-CO1 before analysis). This network prediction analysis revealed that the biomarkers and other genes were functionally enriched in multiple biological processes, including aerobic electron transport chain, aerobic respiration, and mitochondrial ATP synthesis coupled electron transport (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). This part of the results was provided as valuable clues for an in-depth understanding of the potential mechanisms of action of these biomarkers in CO.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Localization analysis and GSEA of biomarkers\u003c/h2\u003e \u003cp\u003eThe results of chromosomal localization analysis showed that PLEK and NIN were localized on chromosomes 2 and 14, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea), which laid a foundation for studying their genetic regulatory characteristics. Furthermore, since COX1 is mitochondrial DNA, it could not be localized on chromosomes. Results of subcellular localization showed that NIN was mainly localized in the nucleus, while PLEK and COX1 were mainly distributed in the cytoplasm (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb). This indicated that these biomarkers might exert their respective functions in different regions within the cell. To further explore the relevant signaling pathways and biological mechanisms involved in the occurrence and development of CO by these biomarkers, GSEA was performed in this study. It was found in the study that PLEK was mainly enriched in 1480 pathways, including \"The transfer of electrons from NADH to ubiquinone that occurs during oxidative phosphorylation\" and \"The transfer of electrons through a series of electron donors and acceptors, generating energy that is ultimately used for synthesis of ATP\", and so on(Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec, \u003cb\u003eSupplementary Table S9\u003c/b\u003e); NIN was mainly enriched in 1403 pathways, including \"The transfer of electrons from NADH to ubiquinone that occurs during oxidative phosphorylation\" and \"The transfer of electrons through a series of electron donors and acceptors, generating energy that is ultimately used for synthesis of ATP\", and so on(Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed, \u003cb\u003eSupplementary Table S10\u003c/b\u003e); COX1 was mainly enriched in 1369 pathways, including \"The transfer of electrons through a series of electron donors and acceptors, generating energy that is ultimately used for synthesis of ATP\" and \"The phosphorylation of ADP to ATP that accompanies the oxidation of a metabolite through the operation of the respiratory chain. Oxidation of compounds establishes a proton gradient across the membrane, providing the energy for ATP synthesis\", and so on (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee, \u003cb\u003eSupplementary Table S11\u003c/b\u003e). For these pathways, the |NES| were all greater than 1, FDRs were all less than 0.25, and the P values were all less than 0.05. These results not only clarified the chromosomal localization and subcellular distribution of PLEK, NIN and COX1, but also revealed the biological processes and regulatory pathways in which PLEK, NIN and COX1 were potentially involved.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Immune infiltration analysis\u003c/h2\u003e \u003cp\u003eTo further analyze the differences in the immune microenvironment between the CO group and the control group, analysis of immune cell infiltration characteristics of the samples was performed in this study. Results showed that the infiltration abundances of 28 immune cells exhibited significant differences between the two groups. Among them, cells such as activated CD4 T cells, activated CD8 T cells, activated dendritic cell, T follicular helper cells, and type 1 T helper cells had lower infiltration levels in the control group, while showing higher infiltration abundances in the CO group (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). Further analysis using the Wilcoxon rank-sum test identified 18 immune cells with statistically significant differences between the two groups (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), which were defined as differential immune cells, including activated CD8 T cells, activated dendritic cells, and T follicular helper cells, etc. (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb). Correlation analysis between differential immune cells revealed that the strongest positive correlation existed between macrophages and myeloid-derived suppressor cells (MDSC) (cor\u0026thinsp;=\u0026thinsp;0.95, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec). Results of correlation analysis between biomarkers and differential immune cells showed that COX1 was negatively correlated with 17 differential immune cells, among which the strongest negative correlation was observed with effector memory CD8 T cell (cor=-0.80, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ed); NIN was positively correlated with 13 differential immune cells, with the highest positive correlations with gamma delta T cell and regulatory T cell (cor\u0026thinsp;=\u0026thinsp;0.67, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ed); PLEK was positively correlated with 15 differential immune cells, among which the strongest positive correlation was with regulatory T cell (cor\u0026thinsp;=\u0026thinsp;0.90, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ed). These results not only confirmed the important roles of biomarkers PLEK, NIN, and COX1 in immune regulation and deepened the understanding of the immunopathological mechanism of CO, but also provided a theoretical basis for the future development of therapeutic strategies targeting specific immune cell subsets.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.6 Molecular regulatory network analysis and drug prediction analysis of biomarkers\u003c/h2\u003e \u003cp\u003ePotential regulatory mechanisms of the biomarkers were further revealed by molecular regulatory network analysis. Results showed that PLEK and NIN were targeted by 16 and 2 miRNAs, respectively, while no targeting miRNAs were predicted for COX1 and its standard symbol MT-CO. Furthermore, a total of 63 LncRNAs were predicted for the two miRNAs targeting PLEK (hsa-miR-141-3p and hsa-miR-200a-3p) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea). To explore potential drugs for treating CO, results of drug prediction analysis showed that 15, 12, and 15 potential targeted drugs were predicted for COX1, NIN, and PLEK, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eb, \u003cb\u003eSupplementary Tables S12-14\u003c/b\u003e). These results were provided as important theoretical bases and resource clues for an in-depth understanding of the molecular regulatory networks (including miRNA and LncRNA levels) of PLEK, NIN, and COX1 in the pathological process of CO, as well as for screening potential therapeutic drugs.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.7 RT-qPCR\u003c/h2\u003e \u003cp\u003eFinally, RT-qPCR experiments were used to verify the expression levels of the 3 biomarkers. The results of the RT-qPCR assay indicated that PLEK and COX1 were found to have significant differences between the control group and the CO group (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Compared with the control group, the expression of PLEK in the CO group was significantly increased, while the expression of COX1 in the CO group was significantly decreased (Figs.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003ea-c). This indicated that the expression changes of these two genes were potentially associated with the occurrence and development of CO.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"3. Discussion","content":"\u003cp\u003eCO is a pathological process with multifactorial causes, which is characterized by an excessive accumulation of body fat and is frequently accompanied by metabolic disorders. Disrupted BCAAs metabolism promotes CO by triggering mitochondrial dysfunction, which adversely affects the sleep cycle, the browning of WAT, and adipocyte-macrophage functionality[\u003cspan additionalcitationids=\"CR6\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Impaired mitochondrial BCAA nitrogen flux in brown adipocytes (BAT) can promote IR[\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Here, we identified three hub genes as key biomarkers and showed that a nomogram model integrating them offers robust predictive power. A comprehensive investigation\u0026mdash;encompassing functional enrichment, localization, regulatory networks, and drug prediction\u0026mdash;coupled with RT-qPCR validation, elucidated their regulatory roles. To our knowledge, this is the first study to explore BCAA metabolism-related biomarkers in CO. These findings provide novel insights for the clinical diagnosis and treatment of CO.\u003c/p\u003e \u003cp\u003eCombining WGCNA, limma difference analysis, and machine learning, we identified three hub genes as biomarkers of CO, including PLEK, NIN, and COX1. PLEK, also known as P47, is a substrate for protein kinase C in platelets and leukocytes[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e] and is related to various autoimmune and inflammatory diseases[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. A recent study demonstrated that PLEK acts as a hub gene via the MAPK and PI3K-Akt signaling pathways in cuprizone-induced demyelination and cognitive impairment in mice[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Over the past decade, researchers have gained a deeper understanding of the PI3K-Akt signaling pathway, which plays a vital role in metabolic diseases[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Modulation of the PI3K-AKT signaling pathway and its downstream molecules represents a potential therapeutic strategy for the treatment of obesity and type 2 diabetes[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. The study on chronic periodontitis demonstrated that the p38 MAPK inhibitor significantly reduced pleckstrin levels induced by IL-1β and LPS, suggesting that PLEK may regulate inflammation via the p38 MAPK pathway[\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Studies have shown that LPS levels are positively correlated with obesity, and LPS-triggered inflammation enhances the degree of obesity[\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Hypoxic adipocytes recruit and activate immune cells, releasing pro-inflammatory cytokines including IL-1β and TNF-α. The sustained increase in LPS and IL-1β leads to chronic activation of the p38 MAPK pathway across multiple cell types\u0026mdash;a key mechanism underlying obesity-related insulin resistance and metabolic syndrome[\u003cspan additionalcitationids=\"CR17\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Together, these findings suggest that PLEK may be upregulated by LPS and IL-1β via the p38 MAPK pathway and contributes critically to CO pathogenesis. The mammalian target of rapamycin (mTOR) signaling is associated with numerous cellular processes such as protein synthesis, cellular metabolism, and growth[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Activation of mTOR signaling can reduce food intake[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. The study showed that ventricular zone expressed pleckstrin homology domain-containing 1 (VEPH1) acts to inhibit mTORC1 signaling through a mechanism involving enhanced TSC1/TSC2 binding, facilitated translocation of TSC2 to the membrane, and a resultant increase in TSC2 Ser1387 phosphorylation[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. It has been demonstrated that leucine induces the dephosphorylation of Sestrin2, leading to its dissociation from GTPase-activating proteins towards Rags 2 (GATOR2) through this mechanism, thereby activating mTORC1 signaling[\u003cspan additionalcitationids=\"CR23 CR24\" citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. Our findings propose a mechanistic hypothesis whereby leucine-induced mTORC1 activation upregulates PLEK expression, thereby exacerbating inflammation and insulin resistance.\u003c/p\u003e \u003cp\u003eObesity-induced insulin resistance is closely related to adipose tissue inflammation, which is driven by adipose tissue macrophages[\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. MIN, also called as ninein, plays a crucial role in macrophages phagocytosis[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. Branched-chain amino acid transaminase 1 (BCAT1), a key rate-limiting enzyme in BCAA metabolism, is highly expressed in macrophages, and BCAT1-mediated metabolic reprogramming is closely associated with inflammatory activation[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. In this study, MIN expression was significantly elevated in CO. Therefore, a hypothese, MIN and BCAAs metabolism regulate the functional status of adipose tissue macrophages, affecting the occurrence and development of chronic inflammation, and thus participating in the pathological process of obesity and its related metabolic complications, maybe reasonable.\u003c/p\u003e \u003cp\u003eResearch has found that the expression of mitochondrial-related pathway genes is downregulated in the subcutaneous adipocytes of obese individuals, accompanied by a decrease in the mitochondrial DNA transcript COX1, which is associated with reduced mitochondrial function, while the expression of inflammatory pathway genes is upregulated[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. The acetyl-CoA derived from BCAA catabolism enters the TCA cycle. The resulting NADH and FADH₂ fuel the electron transport chain, where COX (Complex IV) and other complexes participate in oxidative phosphorylation to generate ATP[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. However, dysfunction of COX1 impedes this entire oxidative phosphorylation process, leading to an accumulation of upstream metabolites, including BCAA breakdown products[\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. These accumulated metabolites are closely associated with obesity and insulin resistance and, in turn, inhibit further BCAA degradation, thereby promoting the occurrence of CO. In conclusion, our study suggests that these three central genes may play a key role in CO.\u003c/p\u003e \u003cp\u003eThe transfer of electrons from NADH to ubiquinone, catalyzed by mitochondrial complex I during oxidative phosphorylation, is a crucial step in mitochondrial electron transport. The elevated NADH levels resulting from mitochondrial complex I dysfunction can lead to insufficient ATP synthesis[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e], which may result in increased food intake, exacerbating obesity. An imblance in the NAD+/NADH is also a common occurence in obesity, insulin resistance and diabetes[\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. To our knowledge, BCAAs catabolism depends on NAD+. The imblance in the NAD+/NADH can inhibit BCAAs catablism, leading to BCAAs accumulation, which further promotes lipogenesis, insulin resistance, and inflammation. The GSEA pathway enrichment analysis showed that the differentially expressed genes were mainly enriched in Mitochondrial electron transport. That suggests they may be important regulatory factors in obesity.\u003c/p\u003e \u003cp\u003eRegulatory T cells (Tregs), a specialized subset of CD4\u0026thinsp;+\u0026thinsp;T cells, are crucial for suppressing immune responses, maintaining tolerance, and preventing autoimmunity[\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. Recently, it has been highlighted that the number of Tregs in adipose tissue of obese individuals is decreased, and this is attributed to insulin resistance[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. Our immune infiltration analysis showed that PLEK and MIN were most strongly and positively correlated with Tregs. These findings suggest that modulating Tregs may be a potential therapeutic strategy for treating obesity-related metabolic disorders.\u003c/p\u003e \u003cp\u003eAlthough multiple weight-loss medications are currently available, their safety in pediatric populations remains undetermined. Therefore, there is an urgent need to explore potential medications. To screen for potential therapeutic drugs for CO, we conducted molecular regulatory network analysis and drug prediction analysis. MALAT1, one of the long noncoding RNAs (lncRNAs) predicted through molecular regulatory network analysis, is highly conserved. Recent research demonstrated that MALAT1 SNP rs3200401 is associated with the risk of CO in Russian populations[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e]. Moreover, the expression of MALAT1 was positively correlated with BMI and other metabolic syndrome-related parameters, including HOMA-IR, total cholesterol and low-density lipoprotein cholesterol[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. Endothelial dysfunction can lead to an increased risk of cadiovascular complication in Obese. Exercise exerted a positive effect against endothelial dysfunction in obese children and adolescents by downregulating MALAT1 expression[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e]. MALAT1 may be a potential therapeutic target.\u003c/p\u003e \u003cp\u003eOur drug prediction analysis also showed that 15, 12, and 15 potential targeted drugs were predicted for COX1, NIN, and PLEK, respectively. Quercetin is a kind of natural flavonoid found, has a wide range of physiology effects, such as anti-flammatory, antioxidant, antidiabetic and anticancer functions[\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. Quercetin is being explored as a treatment for obesity. Studies have revealed that quercetin alleviates obesity by regulating gut microbiota[\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e], which elevates IPA level to activate AhR/IL-22 pathway[\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Another study showed that quercetin reverses obesity through upregulating the expression of 12α-hydroxylase (CYP8B1), thereby facilitating cholesterol conversion to cholic acid[\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e]. Taken together, quercetin holds a siginificant potential therapeutic value in the treatment of obesity. Further research could explore the mechanisms, providing more effective strategies for the treatment of obesity.\u003c/p\u003e \u003cp\u003eCOX1 downregulation in CO suggests impaired mitochondrial respiration, reducing ATP production and disrupting energy homeostasis, thereby promoting fat accumulation. Concurrently, elevated PLEK indicates immune activation, with macrophage infiltration and increased secretion of pro-inflammatory cytokines (e.g., TNF-α, IL-6), exacerbating chronic inflammation in adipose tissue. Our study demonstrated that the expression of COX1 is downregulated in CO, suggesting a potentially significant impairment of mitochondrial respiratory function, which in turn reduces ATP production. This reduction can promote energy storage and inhibit catabolic processes, thus exacerbating the progression of obesity. In contrast, PLEK expression is elevated in the CO group, which may suggest the activation and infiltration of immune cells, particularly macrophages, into adipose tissue. M1-type macrophages secrete potent pro-inflammatory cytokines such as TNF-α, IL-6, and MCP-1, which directly intensify the chronic low-grade inflammation in adipose tissue.\u003c/p\u003e \u003cp\u003eBased on bioinformatics analysis and machine learning, we systematically identified three related candidate genes (PLEK, MIN and COX1). This study may facilitate the exploration of molecular mechanisms in CO, particularly regarding the immune response and drug action. The screened genes could be used for clinical diagnosis and treatment. However, the use of a bioinformatics approach to explore the molecular mechanism and predict potential therapeutic compounds has limitations, and further lab studies are needed to support these results.\u003c/p\u003e"},{"header":"4. Methods","content":"\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Data source\u003c/h2\u003e \u003cp\u003eTwo training sets (GSE205668 and GSE87493) related to CO in this study were obtained from the Gene Expression Omnibus (GEO) database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/geo/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/geo/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The training set GSE205668 (sequencing type: high-throughput; sequencing platform: GPL16791; download time: May 28th, 2025) contained 26 subcutaneous adipose tissue samples from obese children (CO group) and 35 subcutaneous adipose tissue samples from normal controls (control group). The training set GSE87493 (sequencing type: microarray; sequencing platform: GPL6244; download time: May 28th, 2025) contained 12 blood samples from obese children (CO group) and 20 blood samples from normal controls (control group). In addition, 20 branched-chain amino acid metabolism (BCAAM)-related genes (BCAAM-RGs) were obtained from the molecular signatures database (MSigDB) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.gsea-msigdb.org\u003c/span\u003e\u003cspan address=\"https://www.gsea-msigdb.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) using \"REACTOME_BRANCHED_CHAIN_AMINO_ACID_CATABOLISM\" as the search keyword[\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Differential expression analysis\u003c/h2\u003e \u003cp\u003eTo screen out differentially expressed genes (DEGs) in CO, differential expression analysis was first performed on the samples from the CO group and the control group in the training set GSE205668 using the R package \"DESeq2\" (v 1.48.1)[\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]. The first set of DEGs was identified and named DEGs1. Meanwhile, differential expression analysis was conducted on the samples from the CO group and the control group in the training set GSE87493 using the R package \"limma\" (v 3.58.1)[\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e], and the second set of DEGs was identified and named DEGs2. The screening criteria for both analyses were consistent, with the statistical significance threshold set at P\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and |log\u003csub\u003e2\u003c/sub\u003eFold Change (FC)| \u0026gt; 0.1. Subsequently, the R package \"pheatmap\" (v 1.0.12)[\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e] was used to separately generate heatmaps of the expression levels of DEGs1 and DEGs2 between the CO group and the control group. In the heatmaps, genes were sorted in descending order of |log\u003csub\u003e2\u003c/sub\u003eFC| values, displaying the top 10 upregulated and top 10 downregulated genes with the most significant expression differences. Additionally, the R package \"ggplot2\" (v 3.5.1)[\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e] was employed to draw volcano plots for DEGs1 and DEGs2. Similarly, genes in the volcano plots were sorted in descending order of |log\u003csub\u003e2\u003c/sub\u003eFC| values, and the top 10 upregulated and top 10 downregulated genes with the most significant expression differences were labeled to visually present the overall differential expression patterns.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Weighted gene co-expression network analysis (WGCNA)\u003c/h2\u003e \u003cp\u003eTo investigate the expression module characteristics of branched-chain amino acids metabolism-related genes (BCAAM-RGs) in CO, first, based on all samples from the training set GSE205668, the BCAAM-RGs score for each sample was calculated using the single-sample gene set enrichment analysis (ssGSEA) algorithm in the R package \"GSVA\" (v 1.50.5)[\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e]. Differences in BCAAM-RGs scores between the CO group and the control group were compared via the Wilcoxon rank-sum test (with a significance threshold of P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Subsequently, to identify co-expression module genes closely related to BCAAM, weighted gene co-expression network analysis (WGCNA) was performed based on the BCAAM-RGs scores. Hierarchical clustering analysis was conducted on all samples in the training set GSE205668 using the \"hclust\" function in the R package \"GSVA\" (v 1.50.5)[\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e]; outlier samples were identified and excluded to obtain a high-quality sample set for constructing the co-expression network, and a sample clustering tree was plotted. Next, the \"pickSoftThreshold\" function was used to screen the optimal power value within the range of soft threshold (power) parameters 1\u0026ndash;20, with the requirement that the scale-free network evaluation coefficient satisfy 0.8\u0026thinsp;\u0026le;\u0026thinsp;R\u0026sup2; \u0026le; 0.9. A soft threshold screening plot was drawn to determine the correlation threshold between genes that met the scale-free distribution characteristics. The gene expression adjacency matrix was converted into a topological overlap matrix (TOM), and hierarchical clustering was performed via the \"hierarchicalCluster\" function to construct a gene clustering tree. Co-expression modules were identified using the dynamic tree cut algorithm. Finally, the correlation between each module and the BCAAM-RGs score was analyzed, with the threshold set as |correlation coefficient (cor)| \u0026gt; 0.3 and P\u0026thinsp;\u0026lt;\u0026thinsp;0.05. A module-trait correlation heatmap was plotted to visualize the results. One key module significantly positively correlated with the BCAAM-RGs score, and one significantly negatively correlated was selected, and all genes in these two modules were extracted as BCAAM-related key module genes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e4.4 Acquisition of candidate genes and functional enrichment analysis\u003c/h2\u003e \u003cp\u003eTo screen for core DEGs with consistent expression across datasets, the upregulated DEGs (DEGs1-up) from the training set GSE205668 were intersected with the upregulated DEGs (DEGs2-up) from the training set GSE87493. Meanwhile, the downregulated DEGs (DEGs1-down and DEGs2-down) from the two datasets were intersected. Subsequently, the intersected genes from these two parts were merged to obtain DEGs (DEGs3). Thereafter, to identify genes associated with both CO and BCAAM, the R package \"ggvenn\" (v 0.1.10) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://CRAN.R-project.org/package=ggvenn\u003c/span\u003e\u003cspan address=\"https://CRAN.R-project.org/package=ggvenn\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to intersect DEGs3 with BCAAM-related key module genes, and the resulting candidate genes were obtained for subsequent functional validation. The R package \"clusterProfiler\" (v 4.10.1)[\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e] was employed to perform Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology (GO) enrichment analyses on the candidate genes (adjusted (adj). P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), aiming to explore their biological functions and mechanisms. GO consisted of 3 parts, namely biological process (BP), cellular component (CC), and molecular function (MF). Additionally, the search tool for recurring instances of neighbouring genes (STRING) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://string-db.org/\u003c/span\u003e\u003cspan address=\"https://string-db.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to analyze interactions between proteins encoded by candidate genes, with the threshold set as interaction score\u0026thinsp;\u0026gt;\u0026thinsp;0.4. Finally, the results were visualized by constructing a protein-protein interactions (PPI) network diagram using Cytoscape software (v 3.10.0)[\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e4.5 Acquisition of biomarkers\u003c/h2\u003e \u003cp\u003eTo further screen for biomarkers with potential diagnostic value in CO, least absolute shrinkage and selection operator (LASSO) regression analysis, support vector machine-recursive feature elimination (SVM-RFE) analysis, and random forest (RF) analysis were employed to screen biomarkers. First, LASSO regression analysis was performed on the training set GSE205668 using the R package \"glmnet\" (v 4.1-8)[\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e]. The optimal regularization parameter lambda corresponding to the minimum mean squared error was determined through 5-fold cross-validation, and genes with non-zero regression coefficients were screened to obtain feature gene set 1. Meanwhile, regression coefficient plots and cross-validation error curves were generated. Subsequently, SVM-RFE analysis was conducted on the candidate genes using the R package \"caret\" (v 4.7\u0026ndash;1.1)[\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. With a 5-fold cross-validation and a stepwise decrement strategy (removing 1 feature in each iteration) set, the genes corresponding to the model with the highest cross-validation accuracy were selected as feature gene set 2. Furthermore, RF analysis was performed on the candidate genes using the R package \"randomForest\" (v 4.7\u0026ndash;1.2)[\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. The parameter was set to a decision tree number of ntree\u0026thinsp;=\u0026thinsp;100, and the top 10 genes ranked by importance were screened based on the criterion of IncNodePurity\u0026thinsp;\u0026gt;\u0026thinsp;1 to obtain feature gene set 3. Finally, the intersection of the three feature gene sets was obtained by using the R package \"ggvenn\" (v 0.1.10) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://CRAN.R-project.org/package=ggvenn\u003c/span\u003e\u003cspan address=\"https://CRAN.R-project.org/package=ggvenn\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), and biomarkers were acquired.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003e4.6 Construction and evaluation of a nomogram\u003c/h2\u003e \u003cp\u003eTo evaluate the reliability of biomarkers in predicting CO, a nomogram prediction model based on biomarkers was constructed using the R package \"rms\" (v 8.0\u0026ndash;0) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://cran.r-project.org/web/packages/rms/\u003c/span\u003e\u003cspan address=\"http://cran.r-project.org/web/packages/rms/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) with all samples from the training set GSE205668. The scale range of the line corresponding to each biomarker represents its predicted score interval (Points). The total score (Total Points) is obtained by summing the scores of each biomarker, and a higher total score indicates a higher probability of developing CO. The length of the line reflects the contribution weight of the biomarker to the outcome event. Further, the R package \"rms\" (v 8.0\u0026ndash;0) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://cran.r-project.org/web/packages/rms/\u003c/span\u003e\u003cspan address=\"http://cran.r-project.org/web/packages/rms/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to draw a calibration curve to verify the prediction accuracy of the model. The Hosmer-Lemeshow test (HL test) was applied to judge the consistency between the predicted probability and the actual probability. If the P value of the test is greater than 0.05, it indicates that there is no significant difference between the predicted value and the actual value, suggesting that the nomogram model has high calibration accuracy in predicting the incidence of CO. Subsequently, the nomogram's predictive accuracy was assessed via receiver operating characteristic (ROC) curve analysis, performed using the R package \"pROC\" (v 1.19.0.1) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://xrobin.github.io/pROC/\u003c/span\u003e\u003cspan address=\"https://xrobin.github.io/pROC/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The area under the ROC curve (AUC), ranging from 0 to 1, was employed as a metric of discriminatory power, with values exceeding 0.7 considered indicative of good predictive performance. Decision curve analysis (DCA) is a statistical method for evaluating the net benefit of clinical interventions across different probability thresholds, was conducted using the R package \"rmda\" (v 1.6)[\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e]. The DCA curve was created to visualize the nomogram's net benefit across a spectrum of clinically relevant threshold probabilities, with higher nomogram model curve indicating greater clinical utility and improved model performance.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e4.7 Correlation analysis and functional similarity analysis of biomarkers\u003c/h2\u003e \u003cp\u003eTo analyze the correlation between biomarkers, correlation analysis of the biomarkers was performed using the R package \"psych\" (v 2.4.3)[\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e], with the threshold for significant correlation set as |cor| \u0026gt; 0.3 and P\u0026thinsp;\u0026lt;\u0026thinsp;0.05. The results were visualized by drawing a correlation heatmap using the R package \"ggplot2\" (v 3.5.1)[\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. To further evaluate the functional similarity among biomarkers, the semantic similarity of the biomarkers in the three main ontological categories of Gene Ontology (GO) \u0026mdash; biological process (BP), cellular component (CC), and molecular function (MF) \u0026mdash; was calculated using the R package \"GOSemSim\" (v 2.34.0) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/YuLab-SMU/GOSemSim/issues\u003c/span\u003e\u003cspan address=\"https://github.com/YuLab-SMU/GOSemSim/issues\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The calculated average semantic similarity score was used to measure the level of similarity, and a score greater than 0.5 was set as the threshold for high functional similarity.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e4.8 GeneMANIA analysis and localization analysis of biomarkers\u003c/h2\u003e \u003cp\u003eTo deeply analyze the interaction and functional association between biomarkers and their functionally similar genes, the biomarkers were imported into the GeneMANIA online tool (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://genemania.org/\u003c/span\u003e\u003cspan address=\"http://genemania.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), with the specified analysis species set as \"Homo sapiens\". Information from multiple biological databases, including that on co-expression, co-localization, protein-protein interactions, and co-participation in pathways, was integrated by GeneMANIA. Finally, a gene interaction network diagram was constructed to visually display the complex associations between biomarkers. Further, the \"RCircos\" package (v 1.18.4)[\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e] was employed to perform chromosomal localization analysis and generate circular genome maps. By integrating the genomic location information of biomarkers, a visualized chromosome map was generated to visually show the distribution of each biomarker on the chromosome. Then, explore the subcellular localization of biomarkers in the cells. First of all, from the National Center for Biotechnology Information (NCBI) database, to obtained the corresponding FASTA sequences based on biomarkers. Then, the FASTA sequences were imported into the mRNALocater database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://bio-bigdata.cn/mRNALocater/\u003c/span\u003e\u003cspan address=\"http://bio-bigdata.cn/mRNALocater/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), which predicts the localization probability of RNA molecules in 5 subcellular compartments (nucleus, cytoplasm, ribosome, mitochondria, and exosomes).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e4.9 Gene set enrichment analysis (GSEA) of biomarkers\u003c/h2\u003e \u003cp\u003eTo explore the regulatory pathways or biological functions related to the biomarkers, GSEA was conducted for each biomarker based on the CO group samples and control group samples in the training set GSE205668. The Spearman correlation coefficients between other genes and the biomarkers were calculated respectively using the R package \"stats\" (v 0.1.0)[\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e], and the correlation coefficients were sorted in descending order. The gene set \"c5.go.bp.v2024.1.Hs.symbols.gmt\" from the Molecular Signatures Database (MsigDB) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.gsea-msigdb.org/gsea/msigdb/\u003c/span\u003e\u003cspan address=\"https://www.gsea-msigdb.org/gsea/msigdb/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was adopted as the reference gene set. Subsequently, GSEA was performed using the R package \"clusterProfiler\" (v 4.10.0)[\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e], and the significance thresholds were set as |normalized enrichment score (NES)| \u0026gt; 1, False Discovery Rate (FDR)\u0026thinsp;\u0026lt;\u0026thinsp;0.25 and P\u0026thinsp;\u0026lt;\u0026thinsp;0.05. Finally, significant pathways were selected based on P values, and the top 5 most significant signaling pathways were visualized.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e4.10 Immune infiltration analysis\u003c/h2\u003e \u003cp\u003eCO is closely linked to immune dysregulation, typically manifesting as chronic low-grade inflammation and altered immune cell function[\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e, \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e]. To evaluate differences in immune infiltration levels between the CO group and the control group, the relative infiltration abundances of 28 immune cells were calculated using the ssGSEA algorithm[\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e] from the R package \"GSVA\" (v 2.2.0)[\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e] based on all samples from the training set GSE205668, and visualized using the R package \"ggplot2\" (v. 3.5.1)[\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. Furthermore, differences in the infiltration abundance of each immune cell between the two groups were analyzed using the Wilcoxon rank-sum test. Immune cells meeting the significance criterion (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) were defined as differential immune cells, and box plots of their infiltration proportions were plotted and visualized using the R package \"ggplot2\" (v. 3.5.1)[\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e]. Finally, spearman correlation analyses were performed between differential immune cells, and between biomarkers and differential immune cells, respectively, using the R package \"psych\" (v. 2.4.3)[\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e]. The thresholds for correlation significance were set as |cor| \u0026gt; 0.3 and P\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003e4.11 Molecular regulatory network analysis of biomarkers\u003c/h2\u003e \u003cp\u003eTo further explore the upstream regulatory mechanisms of biomarkers, the molecular regulatory network was constructed based on the competing endogenous RNA (ceRNA) theory. Firstly, microRNAs (miRNAs) interacting with biomarkers (mRNAs) were jointly predicted using the diana_microt (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://dianalab.e-ce.uth.gr/html/dianauniverse/index.php?r=microT_CDS\u003c/span\u003e\u003cspan address=\"https://dianalab.e-ce.uth.gr/html/dianauniverse/index.php?r=microT_CDS\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and targetscan (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.targetscan.org/\u003c/span\u003e\u003cspan address=\"https://www.targetscan.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) databases, and only miRNAs present in the prediction results of both databases were retained as candidate miRNAs. Subsequently, long non-coding RNAs (LncRNAs) interacting with candidate miRNAs were predicted using the StarBase database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003eHttp://starbase.sysu.edu.cn\u003c/span\u003e\u003cspan address=\"http://Http://starbase.sysu.edu.cn\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Finally, the LncRNA\u0026ndash;miRNA\u0026ndash;mRNA regulatory network was constructed using the screened LncRNAs and miRNAs, and biomarkers (mRNAs), and the results were visualized using Cytoscape software (v 3.10.0)[\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec23\" class=\"Section2\"\u003e \u003ch2\u003e4.12 Drug prediction analysis of biomarkers\u003c/h2\u003e \u003cp\u003eTo identify potential drugs for the treatment of CO, drug prediction analyses for each biomarker were performed using the drug signatures database (DsigDB) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://dsigdb.tanlab.org/\u003c/span\u003e\u003cspan address=\"http://dsigdb.tanlab.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Based on the prediction results, the potential therapeutic drug that might interact with the biomarkers were identified. Finally, a drug-biomarker interaction network was constructed and visualized using Cytoscape software (v 3.10.0)[\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003e4.13 Reverse transcription quantitative polymerase chain reaction (RT-qPCR)\u003c/h2\u003e \u003cp\u003eTo validate the expression of biomarkers in clinical samples, RT-qPCR analysis was performed. Specifically, a total of 10 blood samples (5 CO and 5 control) were acquired from the clinic in the Fujian Children\u0026rsquo;s Hospital. Written informed consent was obtained from the parents or legal guardians of all minor participants prior to enrollment in the study. The study had the approval of the Fujian Children\u0026rsquo;s Hospital ethics committee (approval number: 2025ETKLR10009). Total RNA from the 10 tissue samples was extracted with the TRIzol reagent (Ambion, USA) according to the manufacturer's protocol. Then the RNA concentration was tested using NanoPhotometer N50. The cDNA was synthesized by reverse transcription using the SureScript First-Strand cDNA Synthesis Kit, and the reaction was performed with S1000TM Thermal Cycler (Bio-Rad, USA). The sequences of all primers can be found in \u003cb\u003eSupplementary Table S1\u003c/b\u003e. The qPCR assay was performed with CFX Connect Real-time Quantitative Fluorescence PCR Instrument (Bio-Rad, USA) (pre-denaturation at 95℃ for 1 min, denaturation at 95℃ for 20s, annealing at 55℃ for 20s, extension at 72℃ for 30s, a total of 40 cycles). The relative quantification of mRNAs was calculated using the 2\u003csup\u003e\u0026minus;ΔΔCT\u003c/sup\u003e method. The results from the RT-qPCR were exported to Excel, and then imported into Graphpad Prism 5 (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.graphpad.com/\u003c/span\u003e\u003cspan address=\"https://www.graphpad.com/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) for statistical analysis and visualization.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e4.14 Statistical analysis\u003c/h2\u003e \u003cp\u003eThe bioinformatics analysis incorporated R software (v 4.4.3) for statistical processing. Statistical significance was defined as a P\u0026thinsp;\u0026lt;\u0026thinsp;0.05. The Wilcoxon rank sum test was used for between-group comparisons, whereas the t-test was applied to analyze RT-qPCR results across groups.\u003c/p\u003e \u003c/div\u003e"},{"header":"Abbreviations","content":"\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"571\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAbbreviation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eFull term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eadj. P\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eadjusted P value\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eAhR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003earyl hydrocarbon receptor\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eATP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eadenosine triphosphate\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eAUC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003earea under the curve\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eBAT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ebrown adipose tissue\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eBCAA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ebranched-chain amino acid\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eBCAAM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ebranched-chain amino acid metabolism\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eBCAAM-RG\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ebranched-chain amino acid metabolism-related gene\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eBCKA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ebranched-chain keto acid\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eBCAT1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ebranched-chain amino acid transaminase 1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eBCAT2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ebranched-chain amino acid transaminase 2\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eBMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ebody mass index\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eBP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ebiological process\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eCC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ecellular component\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003ecDNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ecomplementary DNA\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eceRNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ecompeting endogenous RNA\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eCO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003echildhood obesity\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003ecor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ecorrelation coefficient\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eDCA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003edecision curve analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eDEG\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003edifferentially expressed gene\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eDsigDB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eDrug Signatures Database\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eFADR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003efalse discovery rate\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eFC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003efold change\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eFDR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003efalse discovery rate\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eGEO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eGene Expression Omnibus\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eGeneMANIA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eGene Multiple Association Network Integration Algorithm\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eGO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eGene Ontology\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eGSVA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003egene set variation analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eGSEA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003egene set enrichment analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eHL test\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eHosmer-Lemeshow test\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eHOMA-IR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ehomeostasis model assessment of insulin resistance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eIFNGR1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003einterferon gamma receptor 1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eIL-1\u0026beta;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003einterleukin-1 beta\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eIL-6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003einterleukin-6\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eIPA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eindole-3-propionic acid\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eIR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003einsulin resistance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eIRS-1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003einsulin receptor substrate-1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eJAK1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eJanus kinase 1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eKEGG\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eKyoto Encyclopedia of Genes and Genomes\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eLASSO\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eleast absolute shrinkage and selection operator\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003elimma\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003elinear models for microarray data\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003elncRNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003elong non-coding RNA\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eLPS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003elipopolysaccharide\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eMAE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003emean absolute error\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eMAPK\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003emitogen-activated protein kinase\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eMDSC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003emyeloid-derived suppressor cell\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eMF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003emolecular function\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003emiRNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003emicroRNA\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003emTOR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003emammalian target of rapamycin\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003emTORC1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003emammalian target of rapamycin complex 1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eMSigDB\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eMolecular Signatures Database\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eNADH\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003enicotinamide adenine dinucleotide, reduced form\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eNES\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003enormalized enrichment score\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003ePPI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eprotein-protein interaction\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003ePPAR\u0026gamma;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eperoxisome proliferator-activated receptor gamma\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003ePRDM16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ePR domain-containing protein 16\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eqPCR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003equantitative polymerase chain reaction\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eRF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003erandom forest\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eROC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ereceiver operating characteristic\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eRT-qPCR\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ereverse transcription quantitative polymerase chain reaction\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eSNP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003esingle nucleotide polymorphism\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003essGSEA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003esingle-sample gene set enrichment analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eSTAT1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003esignal transducer and activator of transcription 1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eSTRING\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eSearch Tool for the Retrieval of Interacting Genes/Proteins\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eSVM-RFE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003esupport vector machine-recursive feature elimination\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eT2DM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003etype 2 diabetes mellitus\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eTCA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003etricarboxylic acid cycle\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eTNF-\u0026alpha;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003etumor necrosis factor-alpha\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eTOM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003etopological overlap matrix\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eTreg\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eregulatory T cell\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eWAT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003ewhite adipose tissue\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 118px;\"\u003e\n \u003cp\u003eWGCNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 454px;\"\u003e\n \u003cp\u003eweighted gene co-expression network analysis\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe would like to express our sincere gratitude to all individuals and organizations who supported and assisted us throughout this research. Special thanks to my family\u0026nbsp;for their unwavering support and encouragement throughout this work. In conclusion, we extend our thanks to everyone who has supported and assisted us along the way. Without your support, this research would not have been possible.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eHuijie Zhang: Conceptualization, Data curation, Validation, Visualization, Writing\u0026ndash;original draft, Writing\u0026ndash;review \u0026amp; editing. Ying Lei: Data curation, Validation, Visualization, Writing\u0026ndash;review \u0026amp; editing. Limei Guan: Conceptualization, Supervision, Writing\u0026ndash;review \u0026amp; editing. Hui Liu: Conceptualization, Project administration, Supervision, Writing\u0026ndash;review \u0026amp; editing. This work currently described has not been published, is not being considered for publication elsewhere, and its publication was approved by all authors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability statement\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets analyzed in this study can be found in the [Gene Expression Omnibus (GEO) database] [http://www.ncbi.nlm.nih.gov/geo/] under accession numbers GSE205668 and GSE87493. The branched-chain amino acid metabolism-related gene set used in this study was obtained from the [Molecular Signatures Database (MSigDB)] [https://www.gsea-msigdb.org/gsea/msigdb/].\u003c/p\u003e\n\u003ch1\u003eEthics approval\u003c/h1\u003e\n\u003cp\u003eThis study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Fujian Children\u0026rsquo;s Hospital ethics committee, and the approval number and date of approval are [2025ETKLR10009] and [2025.02.25]. Informed consent was obtained from the parents or legal guardians of all minor participants included in the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to publish\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch1\u003eConflict of Interest\u003c/h1\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003ch1\u003eFunding\u003c/h1\u003e\n\u003cp\u003eThe research reported in this project was generously supported by [Startup Fund for scientific research, Fujian Medical University] under grant agreement number [2021QH1196]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003ePappachan, J. M., Fernandez, C. J. \u0026amp; Ashraf, A. P. Rising tide: The global surge of type 2 diabetes in children and adolescents demands action now. \u003cem\u003eWorld J Diabetes\u003c/em\u003e.\u003cstrong\u003e 15\u003c/strong\u003e, 797\u0026ndash;809 (2024).\u003c/li\u003e\n\u003cli\u003eDi Cesare, M. et al. The epidemiological burden of obesity in childhood: a worldwide epidemic requiring urgent action. \u003cem\u003eBMC Med\u003c/em\u003e.\u003cstrong\u003e 17\u003c/strong\u003e, 212 (2019).\u003c/li\u003e\n\u003cli\u003eDe Spiegeleer, M. et al. Paediatric obesity: a systematic review and pathway mapping of metabolic alterations underlying early disease processes. \u003cem\u003eMol Med\u003c/em\u003e.\u003cstrong\u003e 27\u003c/strong\u003e, 145 (2021).\u003c/li\u003e\n\u003cli\u003eNewgard, C. B. et al. A branched-chain amino acid-related metabolic signature that differentiates obese and lean humans and contributes to insulin resistance. \u003cem\u003eCell Metab\u003c/em\u003e.\u003cstrong\u003e 9\u003c/strong\u003e, 311\u0026ndash;326 (2009).\u003c/li\u003e\n\u003cli\u003eHuang, H., Chen, H., Yao, Y. \u0026amp; Lou, X. Branched-chain amino acids supplementation induces insulin resistance and pro-inflammatory macrophage polarization via INFGR1/JAK1/STAT1 signal pathway. \u003cem\u003eMol Med\u003c/em\u003e.\u003cstrong\u003e 30\u003c/strong\u003e, 149 (2024).\u003c/li\u003e\n\u003cli\u003eLi, H. \u0026amp; Seugnet, L. Decoding the nexus: branched-chain amino acids and their connection with sleep, circadian rhythms, and cardiometabolic health. \u003cem\u003eNeural Regen Res\u003c/em\u003e.\u003cstrong\u003e 20\u003c/strong\u003e, 1350\u0026ndash;1363 (2025).\u003c/li\u003e\n\u003cli\u003eMa, Q. X. et al. BCAA-BCKA axis regulates WAT browning through acetylation of PRDM16. \u003cem\u003eNat Metab\u003c/em\u003e.\u003cstrong\u003e 4\u003c/strong\u003e, 106\u0026ndash;122 (2022).\u003c/li\u003e\n\u003cli\u003eGumus Balikcioglu, P. et al. Branched-chain \u0026alpha;-keto acids and glutamate/glutamine: Biomarkers of insulin resistance in childhood obesity. \u003cem\u003eEndocrinol Diabetes Metab\u003c/em\u003e.\u003cstrong\u003e 6\u003c/strong\u003e, e388 (2023).\u003c/li\u003e\n\u003cli\u003eVerkerke, A. R. P. et al. BCAA-nitrogen flux in brown fat controls metabolic health independent of thermogenesis. \u003cem\u003eCell\u003c/em\u003e.\u003cstrong\u003e 187\u003c/strong\u003e, 2359\u0026ndash;2374.e2318 (2024).\u003c/li\u003e\n\u003cli\u003eMa, A. D., Brass, L. F. \u0026amp; Abrams, C. S. Pleckstrin associates with plasma membranes and induces the formation of membrane projections: requirements for phosphorylation and the NH2-terminal PH domain. \u003cem\u003eJ Cell Biol\u003c/em\u003e.\u003cstrong\u003e 136\u003c/strong\u003e, 1071\u0026ndash;1079 (1997).\u003c/li\u003e\n\u003cli\u003eAlim, M. A. et al. Pleckstrin Levels Are Increased in Patients with Chronic Periodontitis and Regulated via the MAP Kinase-p38\u0026alpha; Signaling Pathway in Gingival Fibroblasts. \u003cem\u003eFront Immunol\u003c/em\u003e.\u003cstrong\u003e 12\u003c/strong\u003e, 801096 (2021).\u003c/li\u003e\n\u003cli\u003eXu, Z., Wen, C. \u0026amp; Wang, W. Role of MAPK and PI3K-Akt signaling pathways in cuprizone-induced demyelination and cognitive impairment in mice. \u003cem\u003eBehav Brain Res\u003c/em\u003e.\u003cstrong\u003e 458\u003c/strong\u003e, 114755 (2024).\u003c/li\u003e\n\u003cli\u003eHuang, X., Liu, G., Guo, J. \u0026amp; Su, Z. The PI3K/AKT pathway in obesity and type 2 diabetes. \u003cem\u003eInt J Biol Sci\u003c/em\u003e.\u003cstrong\u003e 14\u003c/strong\u003e, 1483\u0026ndash;1496 (2018).\u003c/li\u003e\n\u003cli\u003eWang, K. et al. The Effect of Enteric-Derived Lipopolysaccharides on Obesity. \u003cem\u003eInt J Mol Sci\u003c/em\u003e.\u003cstrong\u003e 25\u003c/strong\u003e (2024).\u003c/li\u003e\n\u003cli\u003eSong, B. et al. Association of the gut microbiome with fecal short-chain fatty acids, lipopolysaccharides, and obesity in young Chinese college students. \u003cem\u003eFront Nutr\u003c/em\u003e.\u003cstrong\u003e 10\u003c/strong\u003e, 1057759 (2023).\u003c/li\u003e\n\u003cli\u003eHerti\u0026scaron; Petek, T., Hom\u0026scaron;ak, E., Svetej, M. \u0026amp; Marčun Varda, N. Systemic Inflammation and Oxidative Stress in Childhood Obesity: Sex Differences in Adiposity Indices and Cardiovascular Risk. \u003cem\u003eBiomedicines\u003c/em\u003e.\u003cstrong\u003e 13\u003c/strong\u003e (2024).\u003c/li\u003e\n\u003cli\u003eGonz\u0026aacute;lez-Dom\u0026iacute;nguez, \u0026Aacute;. et al. Altered Metal Homeostasis Associates with Inflammation, Oxidative Stress, Impaired Glucose Metabolism, and Dyslipidemia in the Crosstalk between Childhood Obesity and Insulin Resistance. \u003cem\u003eAntioxidants (Basel)\u003c/em\u003e.\u003cstrong\u003e 11\u003c/strong\u003e (2022).\u003c/li\u003e\n\u003cli\u003eEst\u0026eacute;banez, B., Huang, C. J., Rivera-Viloria, M., Gonz\u0026aacute;lez-Gallego, J. \u0026amp; Cuevas, M. J. Exercise Outcomes in Childhood Obesity-Related Inflammation and Oxidative Status. \u003cem\u003eFront Nutr\u003c/em\u003e.\u003cstrong\u003e 9\u003c/strong\u003e, 886291 (2022).\u003c/li\u003e\n\u003cli\u003eMart\u0026iacute;nez de Morentin, P. B. et al. Hypothalamic mTOR: the rookie energy sensor. \u003cem\u003eCurr Mol Med\u003c/em\u003e.\u003cstrong\u003e 14\u003c/strong\u003e, 3\u0026ndash;21 (2014).\u003c/li\u003e\n\u003cli\u003eZhang, W. et al. Modulation of food intake by mTOR signalling in the dorsal motor nucleus of the vagus in male rats: focus on ghrelin and nesfatin-1. \u003cem\u003eExp Physiol\u003c/em\u003e.\u003cstrong\u003e 98\u003c/strong\u003e, 1696\u0026ndash;1704 (2013).\u003c/li\u003e\n\u003cli\u003eDong, P. et al. Dampened VEPH1 activates mTORC1 signaling by weakening the TSC1/TSC2 association in hepatocellular carcinoma. \u003cem\u003eJ Hepatol\u003c/em\u003e.\u003cstrong\u003e 73\u003c/strong\u003e, 1446\u0026ndash;1459 (2020).\u003c/li\u003e\n\u003cli\u003eKimball, S. R., Gordon, B. S., Moyer, J. E., Dennis, M. D. \u0026amp; Jefferson, L. S. Leucine induced dephosphorylation of Sestrin2 promotes mTORC1 activation. \u003cem\u003eCell Signal\u003c/em\u003e.\u003cstrong\u003e 28\u003c/strong\u003e, 896\u0026ndash;906 (2016).\u003c/li\u003e\n\u003cli\u003eWu, X. et al. FLCN Maintains the Leucine Level in Lysosome to Stimulate mTORC1. \u003cem\u003ePLoS One\u003c/em\u003e.\u003cstrong\u003e 11\u003c/strong\u003e, e0157100 (2016).\u003c/li\u003e\n\u003cli\u003eChen, J. et al. SAR1B senses leucine levels to regulate mTORC1 signalling. \u003cem\u003eNature\u003c/em\u003e.\u003cstrong\u003e 596\u003c/strong\u003e, 281\u0026ndash;284 (2021).\u003c/li\u003e\n\u003cli\u003eCangelosi, A. L. et al. Zonated leucine sensing by Sestrin-mTORC1 in the liver controls the response to dietary leucine. \u003cem\u003eScience\u003c/em\u003e.\u003cstrong\u003e 377\u003c/strong\u003e, 47\u0026ndash;56 (2022).\u003c/li\u003e\n\u003cli\u003eWu, D., Rawal, K., Eeda, V., Lim, H. Y. \u0026amp; Wang, W. Identification and Sorting of Adipose Inflammatory and Metabolically Activated Macrophages in Diet-Induced Obesity. \u003cem\u003eBio Protoc\u003c/em\u003e.\u003cstrong\u003e 15\u003c/strong\u003e, e5479 (2025).\u003c/li\u003e\n\u003cli\u003eOmer, S., Li, J., Yang, C. X. \u0026amp; Harrison, R. E. Ninein promotes F-actin cup formation and inward phagosome movement during phagocytosis in macrophages. \u003cem\u003eMol Biol Cell\u003c/em\u003e.\u003cstrong\u003e 35\u003c/strong\u003e, ar26 (2024).\u003c/li\u003e\n\u003cli\u003ePapathanassiu, A. E. et al. BCAT1 controls metabolic reprogramming in activated human macrophages and is associated with inflammatory diseases. \u003cem\u003eNat Commun\u003c/em\u003e.\u003cstrong\u003e 8\u003c/strong\u003e, 16040 (2017).\u003c/li\u003e\n\u003cli\u003eHeinonen, S. et al. Mitochondria-related transcriptional signature is downregulated in adipocytes in obesity: a study of young healthy MZ twins. \u003cem\u003eDiabetologia\u003c/em\u003e.\u003cstrong\u003e 60\u003c/strong\u003e, 169\u0026ndash;181 (2017).\u003c/li\u003e\n\u003cli\u003eGladyck, S., Aras, S., H\u0026uuml;ttemann, M. \u0026amp; Grossman, L. I. Regulation of COX Assembly and Function by Twin CX(9)C Proteins-Implications for Human Disease. \u003cem\u003eCells\u003c/em\u003e.\u003cstrong\u003e 10\u003c/strong\u003e (2021).\u003c/li\u003e\n\u003cli\u003eWu, G. J. et al. Genistein Triggers Translocation of Estrogen Receptor-Alpha in Mitochondria to Induce Expressions of ATP Synthesis-Associated Genes and Improves Energy Production and Osteoblast Maturation. \u003cem\u003eAm J Chin Med\u003c/em\u003e.\u003cstrong\u003e 49\u003c/strong\u003e, 901\u0026ndash;923 (2021).\u003c/li\u003e\n\u003cli\u003eAdlimoghaddam, A. \u0026amp; Albensi, B. C. The nuclear factor kappa B (NF-\u0026kappa;B) signaling pathway is involved in ammonia-induced mitochondrial dysfunction. \u003cem\u003eMitochondrion\u003c/em\u003e.\u003cstrong\u003e 57\u003c/strong\u003e, 63\u0026ndash;75 (2021).\u003c/li\u003e\n\u003cli\u003eBae, H. R. et al. D-Allulose Ameliorates Dysregulated Macrophage Function and Mitochondrial NADH Homeostasis, Mitigating Obesity-Induced Insulin Resistance. \u003cem\u003eNutrients\u003c/em\u003e.\u003cstrong\u003e 15\u003c/strong\u003e (2023).\u003c/li\u003e\n\u003cli\u003eChen, X. et al. Differential roles of human CD4(+) and CD8(+) regulatory T cells in controlling self-reactive immune responses. \u003cem\u003eNat Immunol\u003c/em\u003e.\u003cstrong\u003e 26\u003c/strong\u003e, 230\u0026ndash;239 (2025).\u003c/li\u003e\n\u003cli\u003eBradley, D. et al. Interferon gamma mediates the reduction of adipose tissue regulatory T cells in human obesity. \u003cem\u003eNat Commun\u003c/em\u003e.\u003cstrong\u003e 13\u003c/strong\u003e, 5606 (2022).\u003c/li\u003e\n\u003cli\u003eSoedono, S. et al. Obese visceral adipose dendritic cells downregulate regulatory T cell development through IL-33. \u003cem\u003eFront Immunol\u003c/em\u003e.\u003cstrong\u003e 15\u003c/strong\u003e, 1335651 (2024).\u003c/li\u003e\n\u003cli\u003eShkurat, T. P. et al. The Role of Genetic Variants in the Long Non-Coding RNA Genes MALAT1 and H19 in the Pathogenesis of Childhood Obesity. \u003cem\u003eNoncoding RNA\u003c/em\u003e.\u003cstrong\u003e 9\u003c/strong\u003e (2023).\u003c/li\u003e\n\u003cli\u003eZhao, W., Yin, Y., Cao, H. \u0026amp; Wang, Y. Exercise Improves Endothelial Function via the lncRNA MALAT1/miR-320a Axis in Obese Children and Adolescents. \u003cem\u003eCardiol Res Pract\u003c/em\u003e.\u003cstrong\u003e 2021\u003c/strong\u003e, 8840698 (2021).\u003c/li\u003e\n\u003cli\u003eShams, E. et al. The effect of quercetin on obesity and reproduction through the expression of genes involved in the hypothalamus-pituitary-gonadal axis. \u003cem\u003eJBRA Assist Reprod\u003c/em\u003e.\u003cstrong\u003e 29\u003c/strong\u003e, 211\u0026ndash;218 (2025).\u003c/li\u003e\n\u003cli\u003eLu, J. et al. Quercetin ameliorates obesity and inflammation via microbial metabolite indole-3-propionic acid in high fat diet-induced obese mice. \u003cem\u003eFront Nutr\u003c/em\u003e.\u003cstrong\u003e 12\u003c/strong\u003e, 1574792 (2025).\u003c/li\u003e\n\u003cli\u003eZhao, L. et al. Quercetin Ameliorates Gut Microbiota Dysbiosis That Drives Hypothalamic Damage and Hepatic Lipogenesis in Monosodium Glutamate-Induced Abdominal Obesity. \u003cem\u003eFront Nutr\u003c/em\u003e.\u003cstrong\u003e 8\u003c/strong\u003e, 671353 (2021).\u003c/li\u003e\n\u003cli\u003eLiu, J. et al. Quercetin-Driven Akkermansia Muciniphila Alleviates Obesity by Modulating Bile Acid Metabolism via an ILA/m(6)A/CYP8B1 Signaling. \u003cem\u003eAdv Sci (Weinh)\u003c/em\u003e.\u003cstrong\u003e 12\u003c/strong\u003e, e2412865 (2025).\u003c/li\u003e\n\u003cli\u003eNeinast, M., Murashige, D. \u0026amp; Arany, Z. Branched Chain Amino Acids. \u003cem\u003eAnnu Rev Physiol\u003c/em\u003e.\u003cstrong\u003e 81\u003c/strong\u003e, 139\u0026ndash;164 (2019).\u003c/li\u003e\n\u003cli\u003eLi, X., Fan, K. \u0026amp; Yang, B. Dynamic transcriptomic landscape of myogenesis in Muscovy ducks (Cairina moschata): integrative analysis of hub genes post-hatching. \u003cem\u003eAnim Biosci\u003c/em\u003e.\u003cstrong\u003e 39\u003c/strong\u003e, 250159 (2026).\u003c/li\u003e\n\u003cli\u003eRitchie, M. E. et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. \u003cem\u003eNucleic Acids Res\u003c/em\u003e.\u003cstrong\u003e 43\u003c/strong\u003e, e47 (2015).\u003c/li\u003e\n\u003cli\u003eMohammed, R., Nader, S. M., Hamza, D. A. \u0026amp; Sabry, M. A. Public health implications of multidrugresistant and methicillinresistant Staphylococcus aureus in retail oysters. \u003cem\u003eSci Rep\u003c/em\u003e.\u003cstrong\u003e 15\u003c/strong\u003e, 4496 (2025).\u003c/li\u003e\n\u003cli\u003ePurssell, E., Frood, S. \u0026amp; Sagoo, R. Beyond the labels: Classifying countries by child health outcomes - A cluster analysis of child mortality and child-health data. \u003cem\u003eGlob Health Action\u003c/em\u003e.\u003cstrong\u003e 18\u003c/strong\u003e, 2526315 (2025).\u003c/li\u003e\n\u003cli\u003eYang, H. et al. Molecular targets of neuroplasticity in ischemic stroke: insights from GEO database, single-cell analysis and immune infiltration analysis. \u003cem\u003eFront Aging Neurosci\u003c/em\u003e.\u003cstrong\u003e 17\u003c/strong\u003e, 1561282 (2025).\u003c/li\u003e\n\u003cli\u003eWu, T. et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. \u003cem\u003eInnovation (Camb)\u003c/em\u003e.\u003cstrong\u003e 2\u003c/strong\u003e, 100141 (2021).\u003c/li\u003e\n\u003cli\u003eOtasek, D., Morris, J. H., Bou\u0026ccedil;as, J., Pico, A. R. \u0026amp; Demchak, B. Cytoscape Automation: empowering workflow-based network analysis. \u003cem\u003eGenome Biol\u003c/em\u003e.\u003cstrong\u003e 20\u003c/strong\u003e, 185 (2019).\u003c/li\u003e\n\u003cli\u003eFriedman, J., Hastie, T. \u0026amp; Tibshirani, R. Regularization Paths for Generalized Linear Models via Coordinate Descent. \u003cem\u003eJ Stat Softw\u003c/em\u003e.\u003cstrong\u003e 33\u003c/strong\u003e, 1\u0026ndash;22 (2010).\u003c/li\u003e\n\u003cli\u003eMishra, A., Maiti, R., Jena, M. \u0026amp; Srinivasan, A. Evaluating machine learning algorithms for prediction of treatment response for sleep disturbances in patients with schizophrenia: A post-hoc analysis from a randomized controlled trial. \u003cem\u003ePsychiatr Danub\u003c/em\u003e.\u003cstrong\u003e 37\u003c/strong\u003e, 46\u0026ndash;54 (2025).\u003c/li\u003e\n\u003cli\u003eHull, T. D. et al. Cyclic Di-GMP phosphodiesterases RmdA and RmdB are involved in regulating colony morphology and development in Streptomyces coelicolor. \u003cem\u003eJ Bacteriol\u003c/em\u003e.\u003cstrong\u003e 194\u003c/strong\u003e, 4642\u0026ndash;4651 (2012).\u003c/li\u003e\n\u003cli\u003eKasyanov, E. D., Yakovleva, Y. V., Mudrakova, T. A., Kasyanova, A. A. \u0026amp; Mazo, G. E. [Comorbidity patterns and structure of depressive episodes in patients with bipolar disorder and major depressive disorder]. \u003cem\u003eZh Nevrol Psikhiatr Im S S Korsakova\u003c/em\u003e.\u003cstrong\u003e 123\u003c/strong\u003e, 108\u0026ndash;114 (2023).\u003c/li\u003e\n\u003cli\u003eYang, Z. et al. Dicoumarol sensitizes hepatocellular carcinoma cells to ferroptosis induced by imidazole ketone erastin. \u003cem\u003eFront Immunol\u003c/em\u003e.\u003cstrong\u003e 16\u003c/strong\u003e, 1531874 (2025).\u003c/li\u003e\n\u003cli\u003eYao, Z. et al. Deciphering the multidimensional impact of IGFBP1 expression on cancer prognosis, genetic alterations, and cellular functionality: A comprehensive Pan-cancer analysis. \u003cem\u003eHeliyon\u003c/em\u003e.\u003cstrong\u003e 10\u003c/strong\u003e, e37402 (2024).\u003c/li\u003e\n\u003cli\u003eDai, W. et al. Influence of adipose tissue immune dysfunction on childhood obesity. \u003cem\u003eCytokine Growth Factor Rev\u003c/em\u003e.\u003cstrong\u003e 65\u003c/strong\u003e, 27\u0026ndash;38 (2022).\u003c/li\u003e\n\u003cli\u003eLagou, M. K. \u0026amp; Karagiannis, G. S. Obesity-induced thymic involution and cancer risk. \u003cem\u003eSemin Cancer Biol\u003c/em\u003e.\u003cstrong\u003e 93\u003c/strong\u003e, 3\u0026ndash;19 (2023).\u003c/li\u003e\n\u003cli\u003eQiu, C. et al. Identification and verification of XDH genes in ROS induced oxidative stress response of osteoarthritis based on bioinformatics analysis. \u003cem\u003eSci Rep\u003c/em\u003e.\u003cstrong\u003e 15\u003c/strong\u003e, 29759 (2025).\u003c/li\u003e\n\u003cli\u003eCao, L. et al. Integrative Analysis of Novel Ferroptosis-Related Genes Signatures as Prognostic Biomarkers in Ovarian Cancer. \u003cem\u003eCancer Rep (Hoboken)\u003c/em\u003e.\u003cstrong\u003e 8\u003c/strong\u003e, e70284 (2025).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Childhood obesity, Branched-chain amino acid metabolism, Transcriptomics, Biomarkers, Immunity","lastPublishedDoi":"10.21203/rs.3.rs-9238869/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9238869/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eChildhood obesity (CO) is a complex chronic disease driven by environmental, behavioral, and genetic factors. Increasing evidence suggests that dysregulation of branched-chain amino acid metabolism (BCAAM) contributes to CO development, highlighting the need to identify reliable biomarkers for diagnosis and treatment. In this study, transcriptome data and BCAAM-related genes were obtained from public databases. Differentially expressed genes were intersected with BCAAM-related genes to identify candidates, and machine learning was applied to screen biomarkers and construct a nomogram model. Multi-dimensional analyses, including functional enrichment, immune infiltration, molecular regulatory network, and drug prediction, were further performed. Three biomarkers, PLEK, NIN, and COX1, were identified, and the nomogram based on them showed good predictive performance. These biomarkers were mainly enriched in mitochondrial energy pathways and were significantly associated with multiple differential immune cells. In addition, PLEK and NIN were predicted to interact with several miRNAs, and multiple potential therapeutic drugs targeting these biomarkers were identified. These findings suggest that PLEK, NIN, and COX1 may serve as promising biomarkers and therapeutic targets for CO.\u003c/p\u003e","manuscriptTitle":"Exploring biomarkers related to branched-chain amino acid metabolism in childhood obesity based on transcriptomics and experimental verification","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-05 16:36:07","doi":"10.21203/rs.3.rs-9238869/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2026-05-03T23:18:52+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"177623072309288118453556138019479592621","date":"2026-04-26T14:41:29+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"295414358902667884784362749986334963323","date":"2026-04-26T05:07:17+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-24T17:16:49+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-24T17:15:21+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-17T13:54:09+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-15T20:58:33+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2026-04-15T20:53:12+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"98d80b57-fc4a-4314-a8ad-de2d64c0d886","owner":[],"postedDate":"May 5th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"","date":"2026-05-03T23:18:52+00:00","index":50,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":67506404,"name":"Health sciences/Biomarkers"},{"id":67506405,"name":"Biological sciences/Computational biology and bioinformatics"}],"tags":[],"updatedAt":"2026-05-05T16:36:07+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-05 16:36:07","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9238869","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9238869","identity":"rs-9238869","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}