Integrated tumor tissue metabolomic and transcriptomic approaches for identifying diagnostic biomarkers in endometrial cancer with diabetes mellitus

Integrated tumor tissue metabolomic and transcriptomic approaches for identifying diagnostic biomarkers in endometrial cancer with diabetes mellitus | 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 Research Article Integrated tumor tissue metabolomic and transcriptomic approaches for identifying diagnostic biomarkers in endometrial cancer with diabetes mellitus Wenwen Zhang, Xueou Liu, Min Fu, Shuqin Chen, Jing Tian, Yong Liu, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8035766/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Endometrial cancer (EC) patients with type 2 diabetes mellitus (DM) often exhibit a more aggressive tumor phenotype and poorer prognosis. However, the underlying metabolic and molecular mechanisms remain poorly understood. Objective: This study aimed to identify diagnostic biomarkers and dysregulated pathways in EC patients with diabetes (EC-DM) through integrated metabolomic and transcriptomic analyses of tumor tissues. Methods Tumor tissues from 20 EC patients (10 EC-DM, 10 non-DM) were analyzed. Untargeted metabolomics used LC–HRMS, and transcriptomics used RNA-seq in 12 patients. Differentially expressed metabolites and genes were identified via machine learning and DESeq2. Pathway and multi-omics integration were performed. Results Multivariate analysis revealed distinct metabolic profiles between EC-DM and EC-NDM groups. Machine learning identified robust DEMs including uric acid, D-malic acid, and guanosine 5′-diphosphate. Pathway analysis showed significant enrichment in purine metabolism, citrate cycle, arginine biosynthesis, and glycerophospholipid metabolism in EC-DM. Transcriptomics identified 1,123 DEGs, with enrichment in monocarboxylic acid metabolic process, terpenoid metabolism, and renin–angiotensin system. Integrated gene–metabolite interaction networks revealed key interactions involving “IL6”, “PPARg”, “PHGDH”, and purine metabolites. Conclusion Our study demonstrates that diabetes reprograms tumor metabolism in endometrial cancer, leading to distinct metabolic and transcriptional alterations. We identified potential diagnostic biomarkers and highlighted dysregulated pathways that may underlie the aggressive phenotype of EC-DM. These findings provide insights into the metabolic interplay between diabetes and endometrial cancer and offer candidates for further validation. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Endometrial cancer (EC) is one of the most common gynecological malignancies worldwide, with increasing incidence linked to metabolic disorders such as obesity and type 2 diabetes mellitus(T2DM)[ 1 – 4 ]. Diabetes has been identified as an independent risk factor for EC, contributing to more aggressive tumor behavior and poorer prognosis[ 5 – 8 ]. However, the molecular mechanisms underlying the interplay between diabetes and endometrial carcinogenesis remain poorly understood. Previous studies have highlighted the metabolic alterations associated with diabetes[ 9 – 12 ], yet there remains a critical gap in understanding how these metabolic changes influence the molecular landscape of endometrial cancer. This underscores the necessity for comprehensive investigations that integrate metabolomic and transcriptomic analyses to elucidate the distinct molecular features of endometrial cancer in patients with diabetes, thereby paving the way for more personalized and effective treatment strategies. Recent advancements in metabolomics and transcriptomics have emerged as pivotal methodologies for elucidating the interplay between cancer and metabolic diseases[ 13 , 14 ]. Prior studies have established that diabetes-related conditions, such as hyperinsulinemia[ 15 , 16 ], chronic hyperglycemia[ 17 , 18 ], and oxidative stress[ 19 , 20 ], can alter the energy metabolism of tumor cells through mechanisms including the activation of the PI3K/Akt/mTOR[ 21 ] pathway and the promotion of lactate accumulation[ 22 ]. Furthermore, abnormalities in pathways related to lipid metabolism[ 23 ] and fatty acid β-oxidation[ 24 ] have been identified in the tumor tissues of endometrial cancer (EC) patients, correlating closely with tumor grading. However, existing research is notably limited in several respects. Most investigations have concentrated on a singular omics layer, such as either metabolite[ 25 , 26 ] or gene expression analysis[ 27 , 28 ], thereby lacking a comprehensive exploration of the "gene-metabolite" regulatory networks that could elucidate the critical regulatory nodes through which diabetes influences EC progression. Additionally, the majority of EC-DM-related studies have relied on blood samples or cellular models[ 29 , 30 ], with a conspicuous scarcity of multi-omics analyses directly targeting tumor tissues, which more accurately reflect the pathological state of tumor cells. Moreover, no studies have employed machine learning or other data mining techniques to identify clinically relevant EC-DM-specific biomarkers, thereby failing to address the clinical demand for early detection and prognostic assessment. This research aims to bridge these gaps by integrating metabolomic and transcriptomic analyses of tumor tissues from EC patients with and without diabetes, thereby providing novel insights into the molecular characteristics and potential diagnostic markers associated with this disease. In this study, we employed a comprehensive methodology that integrates untargeted metabolomics and RNA sequencing to analyze tumor tissues from endometrial cancer (EC) patients, specifically comparing those with diabetes (EC-DM) to those without (EC-NDM). This dual approach not only allows for a detailed characterization of the metabolic and transcriptional landscapes associated with these patient groups but also leverages advanced machine learning techniques and pathway enrichment analyses to discern significant alterations in molecular profiles. By identifying key metabolic and gene expression alterations, this research elucidates the complex interplay between metabolic reprogramming and transcriptional changes induced by diabetes in the context of EC. Materials and methods Ethical review This study adhered to the principles of the Helsinki Declaration. It was approved by the Ethics Committee of Tianjin Central Hospital of Obstetrics and Gynecology (Approval No.: 2021KY088), and written informed consent was obtained from all enrolled participants. Study design and population recruitment From May 2018 to November 2020, a total of 20 patients diagnosed with endometrial carcinoma (EC) were recruited, including 10 cases of EC complicated with diabetes mellitus (EC-DM) and 10 cases of EC without diabetes mellitus (EC-NDM). The detailed inclusion and exclusion criteria for EC patients are presented in Table S1 . Tumor tissue specimens were collected from all enrolled patients at the time of recruitment. All samples underwent non-targeted liquid chromatography-high-resolution mass spectrometry metabolomics. Total RNA was extracted from tumor tissues of 6 EC-DM patients and 6 EC-NDM patients for transcriptomic sequencing. The study design is summarised in Fig. S1 . Extraction of Metabolites from Solid Tissue Approximately 100 mg of frozen tissue was weighed and ground into a fine powder under liquid nitrogen. Metabolites were extracted by adding 120 µL of 50% methanol to each sample, followed by vigorous vortexing to ensure thorough mixing, and incubation at room temperature for 10 min. All metabolite extracts were stored at − 80°C until subsequent analysis. LC–MS/MS analysis All samples were analyzed using a UPLC system (SCIEX, UK) equipped with an ACQUITY UPLC T3 column (100 mm × 2.1 mm, 1.8 µm; Waters, UK) for reversed-phase separation. The column was maintained at 35°C with a flow rate of 0.4 mL/min. Mobile phases were solvent A (water with 0.1% formic acid) and solvent B (acetonitrile with 0.1% formic acid). The gradient program was 0–0.5 min, 5% B; 0.5–7 min, 5–100% B; 7–8 min, 100% B; 8–8.1 min, 100–5% B; and 8.1–10 min, 5% B. The injection volume was 4 µL. Metabolite detection was performed on a TripleTOF 5600plus mass spectrometer (SCIEX, UK) in both positive and negative ion modes, with curtain gas 30 psi, ion source gases 1 and 2 at 60 psi, interface heater 650°C, and IonSpray voltage of + 5000 V (positive) or − 4500 V (negative). Data were acquired in information-dependent acquisition mode over an m/z range of 60–1200, with dynamic exclusion of 4 s and mass calibration every 20 samples. A pooled QC sample was injected every 10 samples to monitor instrument stability and reproducibility. Metabolite data preprocessing and annotation Mass spectrometry data were processed in XCMS for peak detection, alignment, retention time correction, secondary grouping, and isotope/adduct annotation. Raw LC–MS files were converted to mzXML format and analyzed using the XCMS and metaX packages in R. Metabolites were annotated by matching exact m/z values to KEGG and HMDB databases, with a mass tolerance of 10 ppm, and further confirmed by isotopic distribution analysis. An in-house database was used to validate metabolite identifications. Statistical analyses were mainly performed using R software (version 4.4.1). Metabolite data were preprocessed by excluding metabolites with more than 80% missing values across all samples or more than 50% missing values in quality control (QC) samples, imputing the remaining missing values using the k -nearest neighbors (KNN) method, and normalizing the data using probabilistic quotient normalization (PQN) to correct for dilution effects. Identification of differentially expressed metabolites (DEMs) via machine learning analysis Principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), pie charts, heatmaps, and radar charts were employed to visualize the distribution of metabolite samples and the characteristics of the dataset. PCA was performed using the stats package in R, while PLS-DA analysis was conducted with the metaX package. Pie charts and heatmaps were generated using the ggplot2 and pheatmap packages, respectively, and radar charts were created with Origin software. To avoid false positives or false negatives caused by differences between ionization modes, metabolites were screened separately in the positive and negative ion modes to select the most representative core differential metabolites. First, metabolites with p 1 were identified based on t -test results. Given that many metabolites are biologically interdependent and highly correlated, to reduce homology, when the correlation coefficient between two metabolites exceeded 0.95, the feature with the largest absolute mean correlation coefficient with the remaining metabolites was removed. Subsequently, all metabolite data were Z-score normalized across samples for each metabolite (mean-centered and scaled to unit variance) to remove scale differences. Based on the homology-reduced dataset (33 metabolites from the negative ion mode and 47 metabolites from the positive ion mode), three machine learning models—Least Absolute Shrinkage and Selection Operator (LASSO), Random Forest (RF), and Support Vector Machine (SVM)—as well as sPLS-DA (Sparse Partial Least Squares Discriminant Analysis) were compared. Model performance was assessed using 5-fold cross-validation. Predictions from all folds were aggregated to generate a confusion matrix, from which accuracy, sensitivity, specificity, the area under the receiver operating characteristic curve (AUROC), and the corresponding 95% confidence interval (CI) were computed. Following metabolite selection, the Venn diagram illustrates the overlap of metabolites identified using four different methods. Metabolites identified by at least one method were defined as core differential metabolites (CDMs) and included in the multi-omics analysis. To visually demonstrate the discriminatory ability of individual metabolites between EC-NDM and EC-DM, receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) was calculated for metabolites identified by at least three methods. Least Absolute Shrinkage and Selection Operator (LASSO) LASSO is a regression technique widely used in high-dimensional data analysis. It applies L1 regularization to the regression coefficients, which enables both feature selection and model regularization. The model was trained using the glmnet package in R, with α = 1 to implement LASSO regularization (L1 regularization), and the model type was set to binomial for binary classification. The optimal regularization parameter λ was selected through 5-fold cross-validation, with the lambda.min value being chosen as the best-performing parameter. DEMs were identified as those with non-zero coefficients, which were deemed to significantly contribute to the classification. Random Forest (RF) RF is an ensemble learning method that performs classification by constructing multiple decision trees. Each tree learns decision rules from the dataset's features (e.g., metabolites), and the final prediction is obtained through majority voting. During the construction of each tree, a random subset of features and samples is selected at each node, promoting diversity between trees and reducing overfitting. Feature importance is evaluated using the Gini index, where lower values indicate purer splits. To optimize model performance, the hyperparameters ntree (number of trees) and mtry (number of features considered at each split) are tuned using the caret package in R. The ntree parameter is set to 500, and the optimal value of mtry is determined using 5-fold cross-validation. For feature selection, the varImp() function is used to calculate the importance of each feature. Support Vector Machine (SVM) Support Vector Machine (SVM) is a classification algorithm. In this study, a linear SVM model was trained using the e1071 package in R. The regularization parameter C was optimized via grid search with the tune.svm() function, and 5-fold cross-validation was employed to determine the optimal value. Based on the best C obtained, feature selection was performed using Recursive Feature Elimination (RFE). In each iteration, the model was trained on the current feature set, feature importance was quantified as the sum of squared weights derived from the linear SVM decision function, and the least important feature was removed. This procedure was repeated until the desired number of features was retained. The final subset of selected features was subsequently used for further analysis. Sparse Partial Least Squares Discriminant Analysis (sPLS-DA) sPLS-DA is a method designed for high-dimensional data classification, integrating Partial Least Squares (PLS) with L1 regularization (sparsity). This approach enables both feature selection and the reduction of overfitting. The model was trained using the mixOmics package, with sparsity applied to select the most relevant features for classification, thereby reducing the dimensionality of the variables. Subsequently, non-zero features from the first principal component were extracted as core variables for further analysis. Metabolite enrichment analysis Over-representation analysis (ORA) was performed using the MetaboAnalyst 6.0 platform, with pathway annotations retrieved from the KEGG and SMPDB databases. Metabolites identified by at least one of the four aforementioned methods were included in the analysis, and pathways with a p-value < 0.05 were considered statistically significant. RNA extraction and sequencing Total RNA was extracted using TRIzol reagent (Thermo Fisher, 15596018) according to the manufacturer’s protocol[ 31 ]. RNA quantity and integrity were assessed with an Agilent Bioanalyzer 2100 using an RNA 6000 Nano LabChip Kit (Agilent, 5067 − 1511), and samples with an RNA integrity number (RIN) > 7.0 were used for library construction. Poly(A) mRNA was purified from 5 µg of total RNA using Dynabeads Oligo(dT) (Thermo Fisher) with two rounds of selection and fragmented at 94°C for 5–7 min using the Magnesium RNA Fragmentation Module (NEB, E6150). First-strand cDNA synthesis was performed with SuperScript™ II (Invitrogen), followed by second-strand synthesis incorporating dUTP. After A-tailing, indexed adapters were ligated, and size selection (~ 300 ± 50 bp) was performed with AMPure XP beads. U-labeled second strands were degraded using heat-labile UDG (NEB, M0280), and libraries were amplified by PCR (95°C for 3 min; 8 cycles of 98°C for 15 s, 60°C for 15 s, 72°C for 30 s; final extension at 72°C for 5 min). Sequencing was carried out on an Illumina NovaSeq 6000 platform (LC-Bio, Hangzhou, China) to generate 2 × 150 bp paired-end reads. Analysis of Differentially Expressed Genes (DEGs) Differential gene expression analysis between the two groups was performed using the DESeq2 package. A total of 1,123 genes with t-test p-value 1 were identified as differentially expressed genes (DEGs) and included in the multi-omics analysis. Principal component analysis (PCA) of the entire transcriptome dataset was performed with the stats package in R to achieve dimensionality reduction, and the clustering and separation between the EC-NDM and EC-DM patient groups were visualized. Volcano plots illustrating the differential expression results were generated using the ggplot2 package. Pathway enrichment analysis was performed using the Metascape web-based platform. Annotation sources included Gene Ontology (GO) Biological Processes, GO Molecular Functions, Reactome Gene Sets, Canonical Pathways, WikiPathways, Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathways, BioCarta Gene Sets, and Hallmark Gene Sets. Hierarchical clustering of gene expression profiles from EC-NDM and EC-DM groups was conducted using the stats package in R, with complete linkage as the clustering method and Euclidean distance as the metric for both rows and columns. Heatmaps were visualized using the pheatmap package in R. Multi-omics analysis Circos plots To visualize the correlations between differentially expressed genes (DEGs) and major metabolites, Circos plots were generated. Correlation analysis was performed using the Pearson correlation coefficient, with a significance threshold of P ≤ 0.01 and |R| ≥ 0.75. Positive correlations were indicated by red lines, and negative correlations by blue lines. The Circos diagrams were created using the circlize package in R. Gene-metabolite interaction network The Gene–Metabolite Interaction Network module in MetaboAnalyst was employed to analyze metabolites and genes included in the multi-omics dataset, aiming to explore and visualize interactions between functionally related metabolites and genes. Associations between chemicals and human genes were obtained from the STITCH database, and only high-confidence interactions were retained to ensure the reliability of the results. Results Untargeted metabolomic profiling and overall differences between EC-NDM and EC-DM High-resolution Orbitrap mass spectrometry equipped with an electrospray ionization (ESI) source was employed to perform untargeted metabolomic profiling in both positive ion (POS) and negative ion (NEG) modes. Because the ionization efficiencies of metabolites vary according to their chemical structures and functional groups, some compounds can only be effectively detected in a specific mode. Therefore, acquisition in both POS and NEG modes increases metabolite coverage and improves the overall comprehensiveness of detection. Principal component analysis (PCA) revealed clear separation between EC-NDM and EC-DM samples in both NEG and POS modes, indicating substantial differences in their global metabolic profiles (Fig. 1 A, B). To further assess the robustness and reliability of the separation, partial least squares discriminant analysis (PLS-DA) models were constructed, and their performance was evaluated using 200-time permutation tests. The results confirmed that no overfitting occurred in either mode (Fig. 1 C, D). A total of 234 metabolites in NEG mode and 312 metabolites in POS mode were identified. Among these, 45 metabolites in NEG mode and 59 metabolites in POS mode were differentially expressed between the two groups according to the criteria of P 1 ( Fig. 1 E, F). Classification based on the Super Class system showed that the differentially expressed metabolites were predominantly assigned to the categories Organic acids and derivatives (55.4% in NEG, 44.9% in POS), Lipids and lipid-like molecules (10.8% in NEG, 25.2% in POS), and Organoheterocyclic compounds (13.3% in NEG, 14.0% in POS) (Fig. 1 C, D). To compare the relative abundance of key differentially expressed metabolites between the two groups, radar plots of the top 10 metabolites ranked by fold change were generated for both NEG and POS modes. Each dot represents the mean log₁₀-transformed abundance of the corresponding metabolite within each group. These metabolites exhibited marked differences in abundance between EC-DM and EC-NDM, suggesting their potential involvement in the pathogenesis and progression of the disease (Fig. 1 G, H). Classification performance of machine learning models Metabolomic profiling of tumor tissues distinguished EC-NDM (n = 10) from EC-DM (n = 10) in both negative-ion and positive-ion modes. Classification performance was assessed using LASSO, RF, and SVM models. All models demonstrated high discriminative ability (Table 1 , Table 2 ). SVM yielded the highest performance in both modes, achieving an accuracy of 0.850 with an AUROC of 0.900 in negative-ion mode and an accuracy of 0.950 with an AUROC of 0.970 in positive-ion mode, underscoring its superior capability in distinguishing EC-NDM from EC-DM. Table 1 Classification performance of models distinguishing EC-NDM from EC-DM based on negative-ion-mode metabolites. Model Accuracy Sensitivity Specificity AUROC (95% CI) LASSO 0.800 1.000 0.600 0.820 (0.582–1.000) RF 0.800 0.800 0.800 0.730 (0.454–1.000) SVM 0.850 0.900 0.800 0.900 (0.722–1.000) Table 2 Classification performance of models distinguishing EC-NDM from EC-DM based on positive-ion-mode metabolites. Model Accuracy Sensitivity Specificity AUROC (95% CI) LASSO 0.850 0.900 0.800 0.860 (0.668–1.000) RF 0.900 0.900 0.900 0.920 (0.778–1.000) SVM 0.950 1.000 0.900 0.970 (0.904–1.000) Machine learning identifies robust diagnostic metabolites In this study, we applied four methods (LASSO, RF, SVM and sPLS-DA) to identify metabolites distinguishing tumor tissues from EC-NDM and EC-DM patients. Figure 2 A presents metabolites that were identified by ≥ 2 methods. Among them, uric acid, D-malic acid, aspartate, perfluorooctanesulfonic acid, acylcarnitine 6:1 and guanosine 5'-diphosphate were consistently selected by all four methods, indicating their strong robustness in differentiating the two groups (Fig. 2 B ) . Figure 2 C shows the ROC curves of the top-ranked metabolites distinguishing the two patient groups, identified through multiple selection methods. All metabolites yielded AUROC values above 0.7, with several exceeding 0.8, demonstrating robust discriminatory performance and highlighting their potential as candidate biomarkers. Fig. S2 summarizes a total of 55 metabolites identified by ≥ 1 methods, which were defined as CDMs and subsequently incorporated into integrative multi-omics analyses. Differential metabolic pathways and sentinel metabolites in EC-DM To further elucidate the metabolic differences between endometrial cancer with diabetes mellitus (EC-DM) and those without diabetes (EC-NDM), we performed pathway enrichment analysis of the differential metabolites. As shown in Fig. 3 A, several key pathways were significantly enriched in EC-DM, including pyrimidine metabolism, histidine metabolism, pyruvate metabolism, the citrate cycle (TCA cycle), and glycerophospholipid metabolism. Notably, pyrimidine metabolism and arginine biosynthesis showed the strongest enrichment signals (–log10 p > 10). Network analysis using SMPDB further highlighted dysregulation in amino acid metabolism (tyrosine, phenylalanine, aspartate, and glutamate), nucleotide metabolism (purine and pyrimidine), and energy-related processes (glycolysis, Warburg effect, mitochondrial electron transport chain) (Fig. 3 B). Consistent with these pathway-level signals, several sentinel metabolites differed significantly (Fig. 3 C–J). Pyruvic acid and citric acid were higher in EC-DM than in EC-NDM (both p = 0.0049), whereas acetaldehyde (p = 0.014) and glycerol-3-phosphate (p = 0.0059) were lower in EC-DM. In the purine axis, xanthine (p = 0.0037), xanthosine (p = 0.008), guanine (p = 0.013), and uric acid (p = 0.0012) were all reduced in EC-DM. Together, these shifts outline an EC-DM pattern of elevated central-carbon intermediates alongside decreased glycerolipid precursor and diminished purine degradation products. Transcriptomic alterations in metabolic pathways between EC–DM and EC–NDM tumors. To assess whether metabolic alterations in tumor tissues were reflected at the transcriptomic level, RNA sequencing was performed on tumor samples from a subset of patients. Differential expression analysis identified 1,123 DEGs that clearly separated the two groups (Fig. 4 A), including 728 upregulated and 395 downregulated genes (Fig. 4 B). The top 30 upregulated and downregulated DEGs are summarized in Table S2 . Enrichment analyses of upregulated and downregulated genes in the EC-NDM and EC-DM groups identified several significant pathways associated with cellular metabolism and endocrine regulation, including the monocarboxylic acid metabolic process, terpenoid metabolic process, and renin–angiotensin system (Fig. 4 C- 4 D; Fig. S3-S4; Table S3 for DEG lists ). In total, 95 differentially expressed genes (DEGs) were linked to these pathways ( Table S4 ). Among them, UGT1A8, UGT1A9, KLK1, and SPINK1 were also among the top 30 upregulated and downregulated DEGs ( Table S2 ). Together, these transcriptomic findings corroborate the metabolomic data, providing convergent evidence that EC-DM tumors are characterized by enhanced carbon metabolism, disrupted redox homeostasis, and altered nucleotide turnover, while also implicating systemic regulatory pathways such as the renin–angiotensin axis in diabetes-associated endometrial cancer progression. The gene–metabolite interaction network revealed potential molecular mechanisms underlying EC-DM. In the cohort with both metabolomic and transcriptomic data, we performed correlation analyses to investigate the associations between DEMs and DEGs. Significant correlations (p ≤ 0.01) were observed in pathways related to the monocarboxylic acid metabolic process, terpenoid metabolic process, and the renin–angiotensin system ( Fig. S5 ). Most of these correlations were positive, indicating coordinated regulation between metabolites and genes involved in cellular respiration and energy metabolism across transcriptomic and metabolomic levels. To further integrate metabolomic and transcriptomic datasets, we constructed a gene–metabolite interaction network based on 1,123 DEGs and 55 core differential metabolites. The final network identified 10 upregulated genes and 7 downregulated genes interacting with 12 differential metabolites (Fig. 5 ). These interactions highlight the close functional links between genes and metabolites and provide novel evidence for understanding the molecular mechanisms driving EC-DM progression. The specific functional annotations of these genes are summarized in Table S5 . Discussion This study represents a comprehensive multi-omics investigation into the metabolic and transcriptomic differences between endometrial cancer patients with and without diabetes. We identified significant alterations in purine metabolism, central carbon metabolism, and immune-related pathways in EC-DM, supported by both metabolite and gene expression data. A prominent feature of EC-DM tumors was the elevation of glycolytic and TCA intermediates, including pyruvic and citric acid (Fig. 3 ). This pattern suggests enhanced glycolytic flux and mitochondrial oxidation, reflecting a hyperglycemia-driven metabolic reprogramming. Chronic hyperglycemia and hyperinsulinemia activate the PI3K/Akt/mTOR axis, which upregulates glycolytic enzymes and promotes the Warburg effect even under normoxic conditions[ 32 – 34 ]. In endometrial tissue, this metabolic plasticity may enable tumor cells to adapt to oxidative stress and sustain rapid proliferation. The observed upregulation of genes associated with monocarboxylic acid metabolism further supports an intensified carbon flux toward energy production and biosynthesis. These metabolic adjustments, while promoting tumor growth, may also increase dependence on specific pathways—offering potential metabolic vulnerabilities for therapeutic targeting. Another defining alteration in EC-DM was the consistent downregulation of purine degradation intermediates such as uric acid, xanthine, and guanine. Uric acid, a key end-product of purine metabolism, functions as a double-edged molecule with both antioxidant and pro-oxidant roles depending on the cellular context. The reduced levels of purine metabolites observed here likely indicate impaired antioxidant buffering capacity under diabetic conditions (Fig. 3 ), where chronic oxidative stress suppresses xanthine oxidase activity and perturbs nucleotide turnover[ 35 , 36 ]. This imbalance may heighten DNA oxidative damage and genomic instability, contributing to tumor aggressiveness in EC-DM. Xanthine oxidoreductase (XOR) has been shown to affect cell growth in tumors through ROS production, suggesting its role in cancer pathology[ 37 ], reinforcing its significance as both a biomarker and potential metabolic target. The gene-metabolite interaction network highlighted cross-talk between metabolic and immune pathways, with IL6, PPARG, and PHGDH emerging as central players. IL6 is a central cytokine in obesity- and diabetes-associated inflammation, known to activate the JAK/STAT3 pathway and promote angiogenesis and immune evasion[ 38 , 39 ]. Upregulation of IL6 in EC-DM tumors likely contributes to the establishment of a pro-inflammatory, pro-tumorigenic microenvironment. In parallel, dysregulation of PPARG, a master regulator of lipid metabolism and insulin sensitivity, suggests reprogramming of fatty acid oxidation and lipid biosynthesis in response to altered glucose–lipid homeostasis. Meanwhile, PHGDH catalyzes the first step in the serine synthesis pathway, providing one-carbon units and NADPH for nucleotide synthesis and redox control[ 40 ]. The coordinated activation of IL6, PPARG, and PHGDH thus delineates a metabolic–immune axis through which diabetes amplifies anabolic metabolism and inflammatory signaling, jointly driving endometrial tumor progression. The identification of robust metabolic biomarkers—including uric acid, D-malic acid, and guanosine diphosphate—with AUROC values exceeding 0.8 underscores the diagnostic potential of tissue-derived metabolomic signatures. Compared with circulating metabolites, tissue-level profiles more accurately reflect tumor-specific reprogramming under the diabetic milieu. These findings hold promise for developing predictive biomarkers to distinguish EC-DM from EC-NDM or to monitor metabolic therapy response. The main limitations of this study include its relatively small sample size and lack of mechanistic validation. However, the use of tumor tissue–based multi-omics and machine learning provides high specificity and internal consistency, compensating for sample number constraints. Future studies should validate these findings in larger cohorts and use experimental models to confirm the functional roles of key biomarkers. In conclusion, our integrated metabolomic and transcriptomic analysis reveals that diabetes significantly alters the metabolic landscape of endometrial cancer, identifying robust biomarkers and dysregulated pathways that may inform future diagnostic and therapeutic strategies. Conclusion Our study demonstrates that endometrial cancer patients with diabetes (EC-DM) exhibit distinct metabolic and transcriptomic profiles compared to those without diabetes (EC-NDM). Key dysregulated pathways include purine metabolism, TCA cycle, amino acid metabolism, and immune response. Machine learning identified robust diagnostic biomarkers such as uric acid, D-malic acid, and guanosine diphosphate. Integrated analysis revealed coordinated changes in genes and metabolites involved in carbon metabolism, redox homeostasis, and the renin-angiotensin system. These findings suggest that diabetes alters the metabolic landscape of endometrial cancer, offering potential biomarkers for diagnosis and insights into therapeutic targeting. Declarations Author Contributions W Zhang : Data curation, Software, Investigation, Writing-original draft, Funding acquisition. X Liu : Data curation, Software, Investigation, Formal analysis. M Fu : Methodology, writing–review and editing. S Chen : Methodology, Formal analysis, Writing - original draft. J Tian : Conceptualization, Project administration, Investigation, Resources. Y Liu :Conceptualization, Project administration, Investigation, Resources. P Qu : Conceptualization, Project administration, Supervision. Funding Statement This work was supported by Tianjin Science and Technology Planning Project (24JCYBJC01650 and 21JCQNJC00230) Data availability The datasets generated and/or analyzed during the current study have been deposited in the OMIX database of the National Genomics Data Center (NGDC), China National Center for Bioinformation, under accession number OMIX013377. Due to regulations concerning human genetic resources, the data are available under controlled access. The metadata of the dataset are publicly accessible at: https://ngdc.cncb.ac.cn/omix/preview/hAVqoNUo. Access to the full datasets can be granted to qualified researchers through the NGDC data access application process. Full public release of the data will be completed after approval by the relevant regulatory authorities. The following publicly available databases were used in this study: Kyoto Encyclopedia of Genes and Genomes (https://www.genome.jp/kegg/), Human Metabolome Database (https://hmdb.ca/), Small Molecule Pathway Database ( https://smpdb.ca/). Consent for publication Not applicable. Competing interests The authors declare no competing interests. References Onstad MA, Schmandt RE, Lu KH. Addressing the Role of Obesity in Endometrial Cancer Risk, Prevention, and Treatment. J Clin Oncol. 2016;34(35):4225–30. Forte M, Cecere SC, Di Napoli M, Ventriglia J, Tambaro R, Rossetti S, Passarelli A, Casartelli C, Rauso M, Alberico G, et al. Endometrial cancer in the elderly: Characteristics, prognostic and risk factors, and treatment options. Crit Rev Oncol Hematol. 2024;204:104533. McVicker L, Cardwell CR, Edge L, McCluggage WG, Quinn D, Wylie J, McMenamin ÚC. Survival outcomes in endometrial cancer patients according to diabetes: a systematic review and meta-analysis. BMC Cancer. 2022;22(1):427. Peeri NC, Bertrand KA, Na R, De Vivo I, Setiawan VW, Seshan VE, Alemany L, Chen Y, Clarke MA, Clendenen T, et al. Understanding risk factors for endometrial cancer in young women. J Natl Cancer Inst. 2025;117(1):76–88. Friberg E, Orsini N, Mantzoros CS, Wolk A. Diabetes mellitus and risk of endometrial cancer: a meta-analysis. Diabetologia. 2007;50(7):1365–74. Zhang ZH, Su PY, Hao JH, Sun YH. The role of preexisting diabetes mellitus on incidence and mortality of endometrial cancer: a meta-analysis of prospective cohort studies. Int J Gynecol Cancer. 2013;23(2):294–303. Luo J, Beresford S, Chen C, Chlebowski R, Garcia L, Kuller L, Regier M, Wactawski-Wende J, Margolis KL. Association between diabetes, diabetes treatment and risk of developing endometrial cancer. Br J Cancer. 2014;111(7):1432–9. Nief CA, Long SE, McCleary TL, Kidd E, Litkouhi B, Howitt BE. Diabetes mellitus complications associated with recurrence of stage I endometrioid endometrial cancer: A single-center retrospective study. Gynecol Oncol. 2024;190:298–306. Esposito K, Chiodini P, Capuano A, Bellastella G, Maiorino MI, Giugliano D. Metabolic syndrome and endometrial cancer: a meta-analysis. Endocrine. 2014;45(1):28–36. Jin Q, Ma RCW. Metabolomics in Diabetes and Diabetic Complications: Insights from Epidemiological Studies. Cells 2021, 10(11). Gogna N, Krishna M, Oommen AM, Dorai K. Investigating correlations in the altered metabolic profiles of obese and diabetic subjects in a South Indian Asian population using an NMR-based metabolomic approach. Mol Biosyst. 2015;11(2):595–606. Xu F, Tavintharan S, Sum CF, Woon K, Lim SC, Ong CN. Metabolic signature shift in type 2 diabetes mellitus revealed by mass spectrometry-based metabolomics. J Clin Endocrinol Metab. 2013;98(6):E1060–1065. Gong S, Huang R, Wang M, Lian F, Wang Q, Liao Z, Fan C. Comprehensive analysis of the metabolomics and transcriptomics uncovers the dysregulated network and potential biomarkers of Triple Negative Breast Cancer. J Transl Med. 2024;22(1):1016. Piell KM, Poulton CC, Stanley CG, Schultz DJ, Klinge CM. Integrated Metabolomics and Transcriptomics Analysis of Anacardic Acid Inhibition of Breast Cancer Cell Viability. Int J Mol Sci 2024, 25(13). Zhang AMY, Xia YH, Lin JSH, Chu KH, Wang WCK, Ruiter TJJ, Yang JCC, Chen N, Chhuor J, Patil S, et al. Hyperinsulinemia acts via acinar insulin receptors to initiate pancreatic cancer by increasing digestive enzyme production and inflammation. Cell Metab. 2023;35(12):2119–e21352115. Gallagher EJ, LeRoith D. Hyperinsulinaemia in cancer. Nat Rev Cancer. 2020;20(11):629–44. Chang SC, Yang WV. Hyperglycemia, tumorigenesis, and chronic inflammation. Crit Rev Oncol Hematol. 2016;108:146–53. Jeong HS, Lee DH, Kim SH, Lee CH, Shin HM, Kim HR, Cho CH. Hyperglycemia-induced oxidative stress promotes tumor metastasis by upregulating vWF expression in endothelial cells through the transcription factor GATA1. Oncogene. 2022;41(11):1634–46. Lior C, Barki D, Halperin C, Iacobuzio-Donahue CA, Kelsen D, Shouval RS. Mapping the tumor stress network reveals dynamic shifts in the stromal oxidative stress response. Cell Rep. 2024;43(5):114236. Qiu C, Tang C, Tang Y, Su K, Chai X, Zhan Z, Niu X, Li J. RGS5(+) lymphatic endothelial cells facilitate metastasis and acquired drug resistance of breast cancer through oxidative stress-sensing mechanism. Drug Resist Updat. 2024;77:101149. Chakraborty S, Balan M, Sabarwal A, Choueiri TK, Pal S. Metabolic reprogramming in renal cancer: Events of a metabolic disease. Biochim Biophys Acta Rev Cancer. 2021;1876(1):188559. Asiri A, Al Qarni A, Bakillah A. The Interlinking Metabolic Association between Type 2 Diabetes Mellitus and Cancer: Molecular Mechanisms and Therapeutic Insights. Diagnostics (Basel) 2024, 14(19). Wang X, Li Y, Hou X, Li J, Ma X. Lipid metabolism reprogramming in endometrial cancer: biological functions and therapeutic implications. Cell Commun Signal. 2024;22(1):436. Razghonova Y, Mika A, Czapiewska M, Stanczak A, Zygowska P, Wydra DG, Sledzinski T, Abacjew-Chmylko A. Endometrial Cancer Is Associated with Altered Metabolism and Composition of Fatty Acids. Int J Mol Sci 2025, 26(7). Kliemann N, Viallon V, Murphy N, Beeken RJ, Rothwell JA, Rinaldi S, Assi N, van Roekel EH, Schmidt JA, Borch KB, et al. Metabolic signatures of greater body size and their associations with risk of colorectal and endometrial cancers in the European Prospective Investigation into Cancer and Nutrition. BMC Med. 2021;19(1):101. Dossus L, Kouloura E, Biessy C, Viallon V, Siskos AP, Dimou N, Rinaldi S, Merritt MA, Allen N, Fortner R, et al. Prospective analysis of circulating metabolites and endometrial cancer risk. Gynecol Oncol. 2021;162(2):475–81. Sidorkiewicz I, Jóźwik M, Buczyńska A, Erol A, Jóźwik M, Moniuszko M, Jarząbek K, Niemira M, Krętowski A. Identification and subsequent validation of transcriptomic signature associated with metabolic status in endometrial cancer. Sci Rep. 2023;13(1):13763. Fan X, Zou X, Liu C, Cheng W, Zhang S, Geng X, Zhu W. MicroRNA expression profile in serum reveals novel diagnostic biomarkers for endometrial cancer. Biosci Rep 2021, 41(6). Knific T, Vouk K, Smrkolj Š, Prehn C, Adamski J, Rižner TL. Models including plasma levels of sphingomyelins and phosphatidylcholines as diagnostic and prognostic biomarkers of endometrial cancer. J Steroid Biochem Mol Biol. 2018;178:312–21. Yan X, Zhao W, Wei J, Yao Y, Sun G, Wang L, Zhang W, Chen S, Zhou W, Zhao H, et al. A serum lipidomics study for the identification of specific biomarkers for endometrial polyps to distinguish them from endometrial cancer or hyperplasia. Int J Cancer. 2022;150(9):1549–59. Tang YT, Huang YY, Zheng L, Qin SH, Xu XP, An TX, Xu Y, Wu YS, Hu XM, Ping BH, et al. Comparison of isolation methods of exosomes and exosomal RNA from cell culture medium and serum. Int J Mol Med. 2017;40(3):834–44. DeBerardinis RJ, Chandel NS. Fundamentals of cancer metabolism. Sci Adv. 2016;2(5):e1600200. Wu S, Zhang H, Gao C, Chen J, Li H, Meng Z, Bai J, Shen Q, Wu H, Yin T. Hyperglycemia Enhances Immunosuppression and Aerobic Glycolysis of Pancreatic Cancer Through Upregulating Bmi1-UPF1-HK2 Pathway. Cell Mol Gastroenterol Hepatol. 2022;14(5):1146–65. Su Y, Luo Y, Zhang P, Lin H, Pu W, Zhang H, Wang H, Hao Y, Xiao Y, Zhang X, et al. Glucose-induced CRL4(COP1)-p53 axis amplifies glycometabolism to drive tumorigenesis. Mol Cell. 2023;83(13):2316–e23312317. Sautin YY, Johnson RJ. Uric acid: the oxidant-antioxidant paradox. Nucleosides Nucleotides Nucleic Acids. 2008;27(6):608–19. Yan S, Zhang P, Xu W, Liu Y, Wang B, Jiang T, Hua C, Wang X, Xu D, Sun B. Serum Uric Acid Increases Risk of Cancer Incidence and Mortality: A Systematic Review and Meta-Analysis. Mediators Inflamm 2015, 2015:764250. Agarwal A, Banerjee A, Banerjee UC. Xanthine oxidoreductase: a journey from purine metabolism to cardiovascular excitation-contraction coupling. Crit Rev Biotechnol. 2011;31(3):264–80. Chan LC, Li CW, Xia W, Hsu JM, Lee HH, Cha JH, Wang HL, Yang WH, Yen EY, Chang WC, et al. IL-6/JAK1 pathway drives PD-L1 Y112 phosphorylation to promote cancer immune evasion. J Clin Invest. 2019;129(8):3324–38. Catar R, Witowski J, Zhu N, Lücht C, Derrac Soria A, Uceda Fernandez J, Chen L, Jones SA, Fielding CA, Rudolf A, et al. IL-6 Trans-Signaling Links Inflammation with Angiogenesis in the Peritoneal Membrane. J Am Soc Nephrol. 2017;28(4):1188–99. Ahmadian M, Suh JM, Hah N, Liddle C, Atkins AR, Downes M, Evans RM. PPARγ signaling and metabolism: the good, the bad and the future. Nat Med. 2013;19(5):557–66. Additional Declarations No competing interests reported. Supplementary Files SupplementaryTablesS1S5.xlsx SupplementaryFiguresS1S5.docx Cite Share Download PDF Status: Posted Version 1 posted 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-8035766","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":573544733,"identity":"d356db51-049c-412c-aa38-d605127f19e2","order_by":0,"name":"Wenwen Zhang","email":"","orcid":"","institution":"Tianjin Central Hospital of Gynecology Obstetrics","correspondingAuthor":false,"prefix":"","firstName":"Wenwen","middleName":"","lastName":"Zhang","suffix":""},{"id":573544734,"identity":"b74bdfc3-5533-4720-8d66-d3fd234fd9d9","order_by":1,"name":"Xueou Liu","email":"","orcid":"","institution":"Tianjin Central Hospital of Gynecology Obstetrics","correspondingAuthor":false,"prefix":"","firstName":"Xueou","middleName":"","lastName":"Liu","suffix":""},{"id":573544736,"identity":"9ca11f02-78b8-48c6-ace7-ad74c0721d69","order_by":2,"name":"Min Fu","email":"","orcid":"","institution":"Tianjin Central Hospital of Gynecology Obstetrics","correspondingAuthor":false,"prefix":"","firstName":"Min","middleName":"","lastName":"Fu","suffix":""},{"id":573544738,"identity":"d2a2610a-201a-485e-b8d1-e691d29e99e7","order_by":3,"name":"Shuqin Chen","email":"","orcid":"","institution":"Tianjin Central Hospital of Gynecology Obstetrics","correspondingAuthor":false,"prefix":"","firstName":"Shuqin","middleName":"","lastName":"Chen","suffix":""},{"id":573544742,"identity":"5adc4d61-eff3-47eb-bb7e-490f39dcca8d","order_by":4,"name":"Jing Tian","email":"","orcid":"","institution":"Tianjin First Center Hospital","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Tian","suffix":""},{"id":573544750,"identity":"2f09b524-785e-40eb-9941-bc5246f1a6fb","order_by":5,"name":"Yong Liu","email":"","orcid":"","institution":"Nankai University","correspondingAuthor":false,"prefix":"","firstName":"Yong","middleName":"","lastName":"Liu","suffix":""},{"id":573544751,"identity":"adcb83f4-9c26-46eb-917c-ed5e3381b218","order_by":6,"name":"Pengpeng Qu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9ElEQVRIiWNgGAWjYJADxsdgipm5gWgtzMYMDAZAipF4LWzSYC0MBLTw3Uhge/BxR20ev3T7teqCij/R/O1ALT8qtuHUInkjgd1w5pnjxZJzzpTdnnHGIHfGYcYGxp4zt3FqMQDaIs3bdixxw42ctNu8bQa5DUAtzIxtBLT8BWrZD9RSDNIynygtjG01iRsk0o8xg7RsIKRF8szDdsPetgPFEjdymKV5zhjnbgRqOYjPL3zHk489+NlWl8c/I/3hZ54Kudx55w8ffPCjArcWhgOMbUDycAIDA48BQhC3erAsG5CsA2phf4BX4SgYBaNgFIxcAACORl4cXTXwUAAAAABJRU5ErkJggg==","orcid":"","institution":"Tianjin Central Hospital of Gynecology Obstetrics","correspondingAuthor":true,"prefix":"","firstName":"Pengpeng","middleName":"","lastName":"Qu","suffix":""}],"badges":[],"createdAt":"2025-11-05 08:08:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8035766/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8035766/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":100395965,"identity":"a4bbd782-a4fa-43af-8a60-635e757aa1db","added_by":"auto","created_at":"2026-01-16 11:39:41","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":104646,"visible":true,"origin":"","legend":"","description":"","filename":"manuscriptV4.docx","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/9bd83ca04606e6ae9122363f.docx"},{"id":100395261,"identity":"3e7a80bf-96ad-4f80-88bf-e66d8ad47e63","added_by":"auto","created_at":"2026-01-16 11:38:34","extension":"json","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9473,"visible":true,"origin":"","legend":"","description":"","filename":"f83c9d7194424dbfa8d037ed56258f52.json","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/496347606308c549efd6ddb3.json"},{"id":100395559,"identity":"b037f629-6ba1-4c75-a643-80cab2fd25a0","added_by":"auto","created_at":"2026-01-16 11:39:07","extension":"docx","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":930088,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFiguresS1S5.docx","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/8e61cea83f5bd964932964a1.docx"},{"id":100395960,"identity":"d4cf43ba-e443-4372-9761-46989a20699b","added_by":"auto","created_at":"2026-01-16 11:39:40","extension":"xlsx","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38141,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTablesS1S5.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/538d1aa3df2cd79a3d153847.xlsx"},{"id":100396046,"identity":"5d0b4a86-3086-4ff6-89bf-b9d2281dcdd6","added_by":"auto","created_at":"2026-01-16 11:39:50","extension":"xml","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":126575,"visible":true,"origin":"","legend":"","description":"","filename":"f83c9d7194424dbfa8d037ed56258f521enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/6958a15ed111d9638c209a9c.xml"},{"id":100395717,"identity":"8df97ed5-0f0d-427e-9c93-8ffb42d637c3","added_by":"auto","created_at":"2026-01-16 11:39:21","extension":"pdf","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1568596,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/ce4cd40e5680dbacdbc08865.pdf"},{"id":100396060,"identity":"08c589f4-6946-4cc5-a3d6-ebb3a8a2194a","added_by":"auto","created_at":"2026-01-16 11:39:51","extension":"pdf","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":809297,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/4c2f3f30c3bf124877ae05e1.pdf"},{"id":100395967,"identity":"3e339adf-3425-4513-af0a-45d3082c254f","added_by":"auto","created_at":"2026-01-16 11:39:41","extension":"pdf","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":562731,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.3.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/53c50b2e686bdb92c6bdc748.pdf"},{"id":100395517,"identity":"64aadb20-9b44-43f2-9d8b-96473249d8b9","added_by":"auto","created_at":"2026-01-16 11:39:05","extension":"pdf","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":20398045,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.4.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/dee152d7206df9d4323b2981.pdf"},{"id":100395657,"identity":"522e19e9-722f-4265-8731-1a57486b147f","added_by":"auto","created_at":"2026-01-16 11:39:17","extension":"pdf","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":470386,"visible":true,"origin":"","legend":"","description":"","filename":"Fig.5.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/0a6460e2b72e5b5773a5bcf1.pdf"},{"id":100396104,"identity":"28490577-83ff-425b-a63a-1f4f9a7b5ce5","added_by":"auto","created_at":"2026-01-16 11:39:58","extension":"xml","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":124812,"visible":true,"origin":"","legend":"","description":"","filename":"f83c9d7194424dbfa8d037ed56258f521structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/bb545610a33d5c2ba7fec5ec.xml"},{"id":100395327,"identity":"233998f0-04e0-4d9f-a637-7ade5b776f55","added_by":"auto","created_at":"2026-01-16 11:38:50","extension":"html","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":138168,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/63dd7cc5ad52df3679547113.html"},{"id":100395625,"identity":"952e7130-c047-46fc-9881-7384acd74932","added_by":"auto","created_at":"2026-01-16 11:39:10","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1171378,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIdentification of differential metabolites between EC-NDM and EC-DM\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA-B.\u003c/strong\u003ePrincipal component analysis (PCA) score plots of metabolomic data. PC1 and PC2 represent the first and second principal components, respectively. Each dot represents one sample, with dark blue indicating the EC-NDM group and orange indicating the EC-DM group. Below are permutation tests of the PLS-DA models in NEG and POS modes. The x-axis represents the number of permutations, and the y-axis represents the R²Y or Q²values, with dashed lines indicating regression lines.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eC-D.\u003c/strong\u003eDistribution of metabolite superclass categories of DEMs in NEG and POS modes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eE-F.\u003c/strong\u003eHeatmaps of differential metabolite expression levels between EC-NDM and EC-DM.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eG-H.\u003c/strong\u003eRadar plots showing representative significantly altered metabolites between the two groups.\u003c/p\u003e","description":"","filename":"fig1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/49548901c85a0de51e73be4a.jpg"},{"id":100395863,"identity":"5d5d33b9-4341-457b-87e7-660fca8b035c","added_by":"auto","created_at":"2026-01-16 11:39:32","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":552208,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDifferential metabolites between EC-NDM and EC-DM identified by LASSO, RF, SVM, and sPLS-DA.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA.\u003c/strong\u003eMetabolites identified by ≥2 methods. Black squares denote metabolites selected by each method. Red upward arrows indicate increased expression levels and blue downward arrows indicate decreased expression levels in the EC-DM group compared with the EC-NDM group.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eB.\u003c/strong\u003eVenn diagram showing the overlap of differential metabolites (DEMs) identified by different methods.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eC.\u003c/strong\u003eROC curves and AUROC values of top-ranked differential metabolites.\u003c/p\u003e","description":"","filename":"fig2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/0f477b4536d4b4f96b1359da.jpg"},{"id":100395415,"identity":"6d7c16c8-6bc1-449a-b5a9-fb49c9c5a9d6","added_by":"auto","created_at":"2026-01-16 11:39:01","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":786874,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMetabolic pathway enrichment analysis of EC–NDM and EC–DM\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMetabolites identified by ≥1 method (\u003cstrong\u003eFig. S2\u003c/strong\u003e) were subjected to pathway enrichment analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA.\u003c/strong\u003eKEGG-based enrichment analysis showing significantly enriched pathways (p \u0026lt; 0.05). The y-axis represents the enrichment ratio (ER).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eB. \u003c/strong\u003eSMPDB-based pathway network analysis. Each node represents a metabolite set; node color indicates the p value, and node size corresponds to fold enrichment. Edges connect two sets if the proportion of shared metabolites exceeds 25% of the total combined metabolites.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eC–J.\u003c/strong\u003eBoxplots showing normalized relative abundances of differential metabolites in tumor tissues from EC–NDM (n=10, dark blue) and EC–DM (n=10, orange). \u003cstrong\u003eC–F.\u003c/strong\u003ePyruvic acid, citric acid, acetaldehyde, and glycerol-3-phosphate, representing metabolites involved in energy metabolism–related pathways. \u003cstrong\u003eG–J. \u003c/strong\u003eXanthine, xanthosine, guanine, and uric acid, representing metabolites in purine metabolism.\u003c/p\u003e\n\u003cp\u003eGroup differences were evaluated using t tests, and statistical significance is indicated in the plots.\u003c/p\u003e","description":"","filename":"fig3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/424200d7a02dfa736a1542b9.jpg"},{"id":100395948,"identity":"7bc8f464-3753-42a3-ad38-dc3cee31f5fd","added_by":"auto","created_at":"2026-01-16 11:39:38","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":808004,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDifferentially expressed genes and pathway enrichment analysis between EC–NDM and EC–DM\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRNA sequencing was performed on EC–NDM (n=6) and EC–DM (n=6) samples, followed by differential expression and pathway enrichment analyses.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eA.\u003c/strong\u003ePrincipal component analysis based on the 1123 DEGs distinguished EC–NDM (dark blue) from EC–DM (orange) patients.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eB. \u003c/strong\u003eVolcano plot of differentially expressed genes (DEGs; p \u0026lt; 0.05, |log₂FC| \u0026gt; 1). Compared with EC–NDM, red dots indicate genes upregulated in EC–DM, and blue dots indicate genes downregulated in EC–DM.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eC–D.\u003c/strong\u003ePathway enrichment analysis of DEGs using Metascape (GO BP, GO MF, Reactome, CP, WP, KEGG, BioCarta, Hallmark). \u003cstrong\u003eC.\u003c/strong\u003e Top 20 upregulated pathways in EC–DM, ranked by p value. \u003cstrong\u003eD.\u003c/strong\u003e Top 20 downregulated pathways in EC–DM, ranked by p value.\u003c/p\u003e","description":"","filename":"fig4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/0ec01abd9ed8460fa794d08b.jpg"},{"id":100395682,"identity":"4b42448d-2784-43a0-b553-a5d7202ff186","added_by":"auto","created_at":"2026-01-16 11:39:19","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":633807,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGene–metabolite interaction network\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eRelated to \u003cstrong\u003eTable S3\u003c/strong\u003e. A gene–metabolite interaction network was generated in MetaboAnalyst by integrating 1,123 DEGs with 55 metabolites identified by machine learning approaches. The network shows interactions between 10 upregulated genes (red circles), 7 downregulated genes (blue circles), and 12 metabolites (black squares). These genes and metabolites are mainly involved in purine metabolism, the tricarboxylic acid (TCA) cycle, amino acid metabolism, lipid metabolism, and immune-related pathways.\u003c/p\u003e","description":"","filename":"fig5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/5eac8628bc3217057b44df52.jpg"},{"id":104401695,"identity":"ce6cb202-e56e-4ef5-b97a-1fceae8828f0","added_by":"auto","created_at":"2026-03-11 12:13:17","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5227750,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/10f68b1f-8740-4dbe-a5e2-0486c87b3755.pdf"},{"id":100395979,"identity":"e1a7af79-7df4-4a2f-ae18-ee1ef0fb1900","added_by":"auto","created_at":"2026-01-16 11:39:43","extension":"xlsx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":38141,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTablesS1S5.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/945d96b70a2988a6f363dd17.xlsx"},{"id":100396071,"identity":"38e9e1d9-382b-4ea5-a620-9790fc2a6241","added_by":"auto","created_at":"2026-01-16 11:39:52","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":930088,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFiguresS1S5.docx","url":"https://assets-eu.researchsquare.com/files/rs-8035766/v1/6130eb72099fc72454c20828.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integrated tumor tissue metabolomic and transcriptomic approaches for identifying diagnostic biomarkers in endometrial cancer with diabetes mellitus","fulltext":[{"header":"Introduction","content":"\u003cp\u003eEndometrial cancer (EC) is one of the most common gynecological malignancies worldwide, with increasing incidence linked to metabolic disorders such as obesity and type 2 diabetes mellitus(T2DM)[\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Diabetes has been identified as an independent risk factor for EC, contributing to more aggressive tumor behavior and poorer prognosis[\u003cspan additionalcitationids=\"CR6 CR7\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. However, the molecular mechanisms underlying the interplay between diabetes and endometrial carcinogenesis remain poorly understood. Previous studies have highlighted the metabolic alterations associated with diabetes[\u003cspan additionalcitationids=\"CR10 CR11\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e], yet there remains a critical gap in understanding how these metabolic changes influence the molecular landscape of endometrial cancer. This underscores the necessity for comprehensive investigations that integrate metabolomic and transcriptomic analyses to elucidate the distinct molecular features of endometrial cancer in patients with diabetes, thereby paving the way for more personalized and effective treatment strategies.\u003c/p\u003e \u003cp\u003eRecent advancements in metabolomics and transcriptomics have emerged as pivotal methodologies for elucidating the interplay between cancer and metabolic diseases[\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Prior studies have established that diabetes-related conditions, such as hyperinsulinemia[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], chronic hyperglycemia[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e], and oxidative stress[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e], can alter the energy metabolism of tumor cells through mechanisms including the activation of the PI3K/Akt/mTOR[\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e] pathway and the promotion of lactate accumulation[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Furthermore, abnormalities in pathways related to lipid metabolism[\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e] and fatty acid β-oxidation[\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] have been identified in the tumor tissues of endometrial cancer (EC) patients, correlating closely with tumor grading. However, existing research is notably limited in several respects. Most investigations have concentrated on a singular omics layer, such as either metabolite[\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e] or gene expression analysis[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e], thereby lacking a comprehensive exploration of the \"gene-metabolite\" regulatory networks that could elucidate the critical regulatory nodes through which diabetes influences EC progression. Additionally, the majority of EC-DM-related studies have relied on blood samples or cellular models[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e], with a conspicuous scarcity of multi-omics analyses directly targeting tumor tissues, which more accurately reflect the pathological state of tumor cells. Moreover, no studies have employed machine learning or other data mining techniques to identify clinically relevant EC-DM-specific biomarkers, thereby failing to address the clinical demand for early detection and prognostic assessment. This research aims to bridge these gaps by integrating metabolomic and transcriptomic analyses of tumor tissues from EC patients with and without diabetes, thereby providing novel insights into the molecular characteristics and potential diagnostic markers associated with this disease.\u003c/p\u003e \u003cp\u003eIn this study, we employed a comprehensive methodology that integrates untargeted metabolomics and RNA sequencing to analyze tumor tissues from endometrial cancer (EC) patients, specifically comparing those with diabetes (EC-DM) to those without (EC-NDM). This dual approach not only allows for a detailed characterization of the metabolic and transcriptional landscapes associated with these patient groups but also leverages advanced machine learning techniques and pathway enrichment analyses to discern significant alterations in molecular profiles. By identifying key metabolic and gene expression alterations, this research elucidates the complex interplay between metabolic reprogramming and transcriptional changes induced by diabetes in the context of EC.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eEthical review\u003c/h2\u003e \u003cp\u003e This study adhered to the principles of the Helsinki Declaration. It was approved by the Ethics Committee of Tianjin Central Hospital of Obstetrics and Gynecology (Approval No.: 2021KY088), and written informed consent was obtained from all enrolled participants.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStudy design and population recruitment\u003c/h3\u003e\n\u003cp\u003eFrom May 2018 to November 2020, a total of 20 patients diagnosed with endometrial carcinoma (EC) were recruited, including 10 cases of EC complicated with diabetes mellitus (EC-DM) and 10 cases of EC without diabetes mellitus (EC-NDM). The detailed inclusion and exclusion criteria for EC patients are presented in \u003cb\u003eTable \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e. Tumor tissue specimens were collected from all enrolled patients at the time of recruitment. All samples underwent non-targeted liquid chromatography-high-resolution mass spectrometry metabolomics. Total RNA was extracted from tumor tissues of 6 EC-DM patients and 6 EC-NDM patients for transcriptomic sequencing. The study design is summarised in \u003cb\u003eFig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e\u003c/b\u003e.\u003c/p\u003e\n\u003ch3\u003eExtraction of Metabolites from Solid Tissue\u003c/h3\u003e\n\u003cp\u003eApproximately 100 mg of frozen tissue was weighed and ground into a fine powder under liquid nitrogen. Metabolites were extracted by adding 120 \u0026micro;L of 50% methanol to each sample, followed by vigorous vortexing to ensure thorough mixing, and incubation at room temperature for 10 min. All metabolite extracts were stored at \u0026minus;\u0026thinsp;80\u0026deg;C until subsequent analysis.\u003c/p\u003e\n\u003ch3\u003eLC–MS/MS analysis\u003c/h3\u003e\n\u003cp\u003eAll samples were analyzed using a UPLC system (SCIEX, UK) equipped with an ACQUITY UPLC T3 column (100 mm \u0026times; 2.1 mm, 1.8 \u0026micro;m; Waters, UK) for reversed-phase separation. The column was maintained at 35\u0026deg;C with a flow rate of 0.4 mL/min. Mobile phases were solvent A (water with 0.1% formic acid) and solvent B (acetonitrile with 0.1% formic acid). The gradient program was 0\u0026ndash;0.5 min, 5% B; 0.5\u0026ndash;7 min, 5\u0026ndash;100% B; 7\u0026ndash;8 min, 100% B; 8\u0026ndash;8.1 min, 100\u0026ndash;5% B; and 8.1\u0026ndash;10 min, 5% B. The injection volume was 4 \u0026micro;L. Metabolite detection was performed on a TripleTOF 5600plus mass spectrometer (SCIEX, UK) in both positive and negative ion modes, with curtain gas 30 psi, ion source gases 1 and 2 at 60 psi, interface heater 650\u0026deg;C, and IonSpray voltage of +\u0026thinsp;5000 V (positive) or \u0026minus;\u0026thinsp;4500 V (negative). Data were acquired in information-dependent acquisition mode over an \u003cem\u003em/z\u003c/em\u003e range of 60\u0026ndash;1200, with dynamic exclusion of 4 s and mass calibration every 20 samples. A pooled QC sample was injected every 10 samples to monitor instrument stability and reproducibility.\u003c/p\u003e\n\u003ch3\u003eMetabolite data preprocessing and annotation\u003c/h3\u003e\n\u003cp\u003eMass spectrometry data were processed in XCMS for peak detection, alignment, retention time correction, secondary grouping, and isotope/adduct annotation. Raw LC\u0026ndash;MS files were converted to mzXML format and analyzed using the XCMS and metaX packages in R. Metabolites were annotated by matching exact \u003cem\u003em/z\u003c/em\u003e values to KEGG and HMDB databases, with a mass tolerance of 10 ppm, and further confirmed by isotopic distribution analysis. An in-house database was used to validate metabolite identifications. Statistical analyses were mainly performed using R software (version 4.4.1). Metabolite data were preprocessed by excluding metabolites with more than 80% missing values across all samples or more than 50% missing values in quality control (QC) samples, imputing the remaining missing values using the \u003cem\u003ek\u003c/em\u003e-nearest neighbors (KNN) method, and normalizing the data using probabilistic quotient normalization (PQN) to correct for dilution effects.\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eIdentification of differentially expressed metabolites (DEMs) via machine learning analysis\u003c/h2\u003e \u003cp\u003ePrincipal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), pie charts, heatmaps, and radar charts were employed to visualize the distribution of metabolite samples and the characteristics of the dataset. PCA was performed using the stats package in R, while PLS-DA analysis was conducted with the metaX package. Pie charts and heatmaps were generated using the ggplot2 and pheatmap packages, respectively, and radar charts were created with Origin software.\u003c/p\u003e \u003cp\u003eTo avoid false positives or false negatives caused by differences between ionization modes, metabolites were screened separately in the positive and negative ion modes to select the most representative core differential metabolites. First, metabolites with \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and |log₂FC| \u0026gt; 1 were identified based on \u003cem\u003et\u003c/em\u003e-test results. Given that many metabolites are biologically interdependent and highly correlated, to reduce homology, when the correlation coefficient between two metabolites exceeded 0.95, the feature with the largest absolute mean correlation coefficient with the remaining metabolites was removed. Subsequently, all metabolite data were Z-score normalized across samples for each metabolite (mean-centered and scaled to unit variance) to remove scale differences.\u003c/p\u003e \u003cp\u003eBased on the homology-reduced dataset (33 metabolites from the negative ion mode and 47 metabolites from the positive ion mode), three machine learning models\u0026mdash;Least Absolute Shrinkage and Selection Operator (LASSO), Random Forest (RF), and Support Vector Machine (SVM)\u0026mdash;as well as sPLS-DA (Sparse Partial Least Squares Discriminant Analysis) were compared. Model performance was assessed using 5-fold cross-validation. Predictions from all folds were aggregated to generate a confusion matrix, from which accuracy, sensitivity, specificity, the area under the receiver operating characteristic curve (AUROC), and the corresponding 95% confidence interval (CI) were computed.\u003c/p\u003e \u003cp\u003eFollowing metabolite selection, the Venn diagram illustrates the overlap of metabolites identified using four different methods. Metabolites identified by at least one method were defined as core differential metabolites (CDMs) and included in the multi-omics analysis. To visually demonstrate the discriminatory ability of individual metabolites between EC-NDM and EC-DM, receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) was calculated for metabolites identified by at least three methods.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eLeast Absolute Shrinkage and Selection Operator (LASSO)\u003c/h3\u003e\n\u003cp\u003eLASSO is a regression technique widely used in high-dimensional data analysis. It applies L1 regularization to the regression coefficients, which enables both feature selection and model regularization. The model was trained using the glmnet package in R, with \u003cem\u003eα\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1 to implement LASSO regularization (L1 regularization), and the model type was set to binomial for binary classification. The optimal regularization parameter \u003cem\u003eλ\u003c/em\u003e was selected through 5-fold cross-validation, with the lambda.min value being chosen as the best-performing parameter. DEMs were identified as those with non-zero coefficients, which were deemed to significantly contribute to the classification.\u003c/p\u003e\n\u003ch3\u003eRandom Forest (RF)\u003c/h3\u003e\n\u003cp\u003eRF is an ensemble learning method that performs classification by constructing multiple decision trees. Each tree learns decision rules from the dataset's features (e.g., metabolites), and the final prediction is obtained through majority voting. During the construction of each tree, a random subset of features and samples is selected at each node, promoting diversity between trees and reducing overfitting. Feature importance is evaluated using the Gini index, where lower values indicate purer splits. To optimize model performance, the hyperparameters ntree (number of trees) and mtry (number of features considered at each split) are tuned using the caret package in R. The ntree parameter is set to 500, and the optimal value of mtry is determined using 5-fold cross-validation. For feature selection, the varImp() function is used to calculate the importance of each feature.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eSupport Vector Machine (SVM)\u003c/h2\u003e \u003cp\u003eSupport Vector Machine (SVM) is a classification algorithm. In this study, a linear SVM model was trained using the e1071 package in R. The regularization parameter \u003cem\u003eC\u003c/em\u003e was optimized via grid search with the tune.svm() function, and 5-fold cross-validation was employed to determine the optimal value. Based on the best \u003cem\u003eC\u003c/em\u003e obtained, feature selection was performed using Recursive Feature Elimination (RFE). In each iteration, the model was trained on the current feature set, feature importance was quantified as the sum of squared weights derived from the linear SVM decision function, and the least important feature was removed. This procedure was repeated until the desired number of features was retained. The final subset of selected features was subsequently used for further analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eSparse Partial Least Squares Discriminant Analysis (sPLS-DA)\u003c/h2\u003e \u003cp\u003esPLS-DA is a method designed for high-dimensional data classification, integrating Partial Least Squares (PLS) with L1 regularization (sparsity). This approach enables both feature selection and the reduction of overfitting. The model was trained using the mixOmics package, with sparsity applied to select the most relevant features for classification, thereby reducing the dimensionality of the variables. Subsequently, non-zero features from the first principal component were extracted as core variables for further analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eMetabolite enrichment analysis\u003c/h2\u003e \u003cp\u003eOver-representation analysis (ORA) was performed using the MetaboAnalyst 6.0 platform, with pathway annotations retrieved from the KEGG and SMPDB databases. Metabolites identified by at least one of the four aforementioned methods were included in the analysis, and pathways with a p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 were considered statistically significant.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eRNA extraction and sequencing\u003c/h2\u003e \u003cp\u003eTotal RNA was extracted using TRIzol reagent (Thermo Fisher, 15596018) according to the manufacturer\u0026rsquo;s protocol[\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. RNA quantity and integrity were assessed with an Agilent Bioanalyzer 2100 using an RNA 6000 Nano LabChip Kit (Agilent, 5067\u0026thinsp;\u0026minus;\u0026thinsp;1511), and samples with an RNA integrity number (RIN)\u0026thinsp;\u0026gt;\u0026thinsp;7.0 were used for library construction. Poly(A) mRNA was purified from 5 \u0026micro;g of total RNA using Dynabeads Oligo(dT) (Thermo Fisher) with two rounds of selection and fragmented at 94\u0026deg;C for 5\u0026ndash;7 min using the Magnesium RNA Fragmentation Module (NEB, E6150). First-strand cDNA synthesis was performed with SuperScript\u0026trade; II (Invitrogen), followed by second-strand synthesis incorporating dUTP. After A-tailing, indexed adapters were ligated, and size selection (~\u0026thinsp;300\u0026thinsp;\u0026plusmn;\u0026thinsp;50 bp) was performed with AMPure XP beads. U-labeled second strands were degraded using heat-labile UDG (NEB, M0280), and libraries were amplified by PCR (95\u0026deg;C for 3 min; 8 cycles of 98\u0026deg;C for 15 s, 60\u0026deg;C for 15 s, 72\u0026deg;C for 30 s; final extension at 72\u0026deg;C for 5 min). Sequencing was carried out on an Illumina NovaSeq 6000 platform (LC-Bio, Hangzhou, China) to generate 2 \u0026times; 150 bp paired-end reads.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eAnalysis of Differentially Expressed Genes (DEGs)\u003c/h2\u003e \u003cp\u003eDifferential gene expression analysis between the two groups was performed using the DESeq2 package. A total of 1,123 genes with t-test p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and absolute log₂ fold change (|log₂FC|)\u0026thinsp;\u0026gt;\u0026thinsp;1 were identified as differentially expressed genes (DEGs) and included in the multi-omics analysis. Principal component analysis (PCA) of the entire transcriptome dataset was performed with the stats package in R to achieve dimensionality reduction, and the clustering and separation between the EC-NDM and EC-DM patient groups were visualized. Volcano plots illustrating the differential expression results were generated using the ggplot2 package. Pathway enrichment analysis was performed using the Metascape web-based platform. Annotation sources included Gene Ontology (GO) Biological Processes, GO Molecular Functions, Reactome Gene Sets, Canonical Pathways, WikiPathways, Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathways, BioCarta Gene Sets, and Hallmark Gene Sets. Hierarchical clustering of gene expression profiles from EC-NDM and EC-DM groups was conducted using the stats package in R, with complete linkage as the clustering method and Euclidean distance as the metric for both rows and columns. Heatmaps were visualized using the pheatmap package in R.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eMulti-omics analysis\u003c/h2\u003e \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e \u003ch2\u003eCircos plots\u003c/h2\u003e \u003cp\u003eTo visualize the correlations between differentially expressed genes (DEGs) and major metabolites, Circos plots were generated. Correlation analysis was performed using the Pearson correlation coefficient, with a significance threshold of \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026le;\u0026thinsp;0.01 and |R| \u0026ge; 0.75. Positive correlations were indicated by red lines, and negative correlations by blue lines. The Circos diagrams were created using the circlize package in R.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eGene-metabolite interaction network\u003c/h2\u003e \u003cp\u003eThe Gene\u0026ndash;Metabolite Interaction Network module in MetaboAnalyst was employed to analyze metabolites and genes included in the multi-omics dataset, aiming to explore and visualize interactions between functionally related metabolites and genes. Associations between chemicals and human genes were obtained from the STITCH database, and only high-confidence interactions were retained to ensure the reliability of the results.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eUntargeted metabolomic profiling and overall differences between EC-NDM and EC-DM\u003c/h2\u003e \u003cp\u003eHigh-resolution Orbitrap mass spectrometry equipped with an electrospray ionization (ESI) source was employed to perform untargeted metabolomic profiling in both positive ion (POS) and negative ion (NEG) modes. Because the ionization efficiencies of metabolites vary according to their chemical structures and functional groups, some compounds can only be effectively detected in a specific mode. Therefore, acquisition in both POS and NEG modes increases metabolite coverage and improves the overall comprehensiveness of detection. Principal component analysis (PCA) revealed clear separation between EC-NDM and EC-DM samples in both NEG and POS modes, indicating substantial differences in their global metabolic profiles (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA, B). To further assess the robustness and reliability of the separation, partial least squares discriminant analysis (PLS-DA) models were constructed, and their performance was evaluated using 200-time permutation tests. The results confirmed that no overfitting occurred in either mode (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC, D). A total of 234 metabolites in NEG mode and 312 metabolites in POS mode were identified. Among these, 45 metabolites in NEG mode and 59 metabolites in POS mode were differentially expressed between the two groups according to the criteria of \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and |log₂FC| \u0026gt; 1 \u003cb\u003e(\u003c/b\u003eFig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eE, F). Classification based on the Super Class system showed that the differentially expressed metabolites were predominantly assigned to the categories Organic acids and derivatives (55.4% in NEG, 44.9% in POS), Lipids and lipid-like molecules (10.8% in NEG, 25.2% in POS), and Organoheterocyclic compounds (13.3% in NEG, 14.0% in POS) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC, D). To compare the relative abundance of key differentially expressed metabolites between the two groups, radar plots of the top 10 metabolites ranked by fold change were generated for both NEG and POS modes. Each dot represents the mean log₁₀-transformed abundance of the corresponding metabolite within each group. These metabolites exhibited marked differences in abundance between EC-DM and EC-NDM, suggesting their potential involvement in the pathogenesis and progression of the disease (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eG, H).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eClassification performance of machine learning models\u003c/h2\u003e \u003cp\u003eMetabolomic profiling of tumor tissues distinguished EC-NDM (n\u0026thinsp;=\u0026thinsp;10) from EC-DM (n\u0026thinsp;=\u0026thinsp;10) in both negative-ion and positive-ion modes. Classification performance was assessed using LASSO, RF, and SVM models. All models demonstrated high discriminative ability (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). SVM yielded the highest performance in both modes, achieving an accuracy of 0.850 with an AUROC of 0.900 in negative-ion mode and an accuracy of 0.950 with an AUROC of 0.970 in positive-ion mode, underscoring its superior capability in distinguishing EC-NDM from EC-DM.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eClassification performance of models distinguishing EC-NDM from EC-DM based on negative-ion-mode metabolites.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAUROC (95% CI)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLASSO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.600\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.820 (0.582\u0026ndash;1.000)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.730 (0.454\u0026ndash;1.000)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.850\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.900\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.900 (0.722\u0026ndash;1.000)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eClassification performance of models distinguishing EC-NDM from EC-DM based on positive-ion-mode metabolites.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAUROC (95% CI)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLASSO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.850\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.900\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.800\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.860 (0.668\u0026ndash;1.000)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRF\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.900\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.900\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.900\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.920 (0.778\u0026ndash;1.000)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.950\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.900\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.970 (0.904\u0026ndash;1.000)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eMachine learning identifies robust diagnostic metabolites\u003c/h2\u003e \u003cp\u003eIn this study, we applied four methods (LASSO, RF, SVM and sPLS-DA) to identify metabolites distinguishing tumor tissues from EC-NDM and EC-DM patients. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA presents metabolites that were identified by \u0026ge;\u0026thinsp;2 methods. Among them, uric acid, D-malic acid, aspartate, perfluorooctanesulfonic acid, acylcarnitine 6:1 and guanosine 5'-diphosphate were consistently selected by all four methods, indicating their strong robustness in differentiating the two groups (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB\u003cb\u003e)\u003c/b\u003e. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC shows the ROC curves of the top-ranked metabolites distinguishing the two patient groups, identified through multiple selection methods. All metabolites yielded AUROC values above 0.7, with several exceeding 0.8, demonstrating robust discriminatory performance and highlighting their potential as candidate biomarkers. \u003cb\u003eFig. \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e\u003c/b\u003e summarizes a total of 55 metabolites identified by \u0026ge;\u0026thinsp;1 methods, which were defined as CDMs and subsequently incorporated into integrative multi-omics analyses.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eDifferential metabolic pathways and sentinel metabolites in EC-DM\u003c/h2\u003e \u003cp\u003eTo further elucidate the metabolic differences between endometrial cancer with diabetes mellitus (EC-DM) and those without diabetes (EC-NDM), we performed pathway enrichment analysis of the differential metabolites. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA, several key pathways were significantly enriched in EC-DM, including pyrimidine metabolism, histidine metabolism, pyruvate metabolism, the citrate cycle (TCA cycle), and glycerophospholipid metabolism. Notably, pyrimidine metabolism and arginine biosynthesis showed the strongest enrichment signals (\u0026ndash;log10 p\u0026thinsp;\u0026gt;\u0026thinsp;10). Network analysis using SMPDB further highlighted dysregulation in amino acid metabolism (tyrosine, phenylalanine, aspartate, and glutamate), nucleotide metabolism (purine and pyrimidine), and energy-related processes (glycolysis, Warburg effect, mitochondrial electron transport chain) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eConsistent with these pathway-level signals, several sentinel metabolites differed significantly (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC\u0026ndash;J). Pyruvic acid and citric acid were higher in EC-DM than in EC-NDM (both p\u0026thinsp;=\u0026thinsp;0.0049), whereas acetaldehyde (p\u0026thinsp;=\u0026thinsp;0.014) and glycerol-3-phosphate (p\u0026thinsp;=\u0026thinsp;0.0059) were lower in EC-DM. In the purine axis, xanthine (p\u0026thinsp;=\u0026thinsp;0.0037), xanthosine (p\u0026thinsp;=\u0026thinsp;0.008), guanine (p\u0026thinsp;=\u0026thinsp;0.013), and uric acid (p\u0026thinsp;=\u0026thinsp;0.0012) were all reduced in EC-DM. Together, these shifts outline an EC-DM pattern of elevated central-carbon intermediates alongside decreased glycerolipid precursor and diminished purine degradation products.\u003c/p\u003e \u003cp\u003e \u003cb\u003eTranscriptomic alterations in metabolic pathways between EC\u0026ndash;DM and EC\u0026ndash;NDM tumors.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eTo assess whether metabolic alterations in tumor tissues were reflected at the transcriptomic level, RNA sequencing was performed on tumor samples from a subset of patients. Differential expression analysis identified 1,123 DEGs that clearly separated the two groups (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA), including 728 upregulated and 395 downregulated genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB). The top 30 upregulated and downregulated DEGs are summarized in \u003cb\u003eTable \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e\u003c/b\u003e. Enrichment analyses of upregulated and downregulated genes in the EC-NDM and EC-DM groups identified several significant pathways associated with cellular metabolism and endocrine regulation, including the monocarboxylic acid metabolic process, terpenoid metabolic process, and renin\u0026ndash;angiotensin system (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC-\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD; \u003cb\u003eFig. S3-S4; Table S3 for DEG lists\u003c/b\u003e). In total, 95 differentially expressed genes (DEGs) were linked to these pathways (\u003cb\u003eTable S4\u003c/b\u003e). Among them, UGT1A8, UGT1A9, KLK1, and SPINK1 were also among the top 30 upregulated and downregulated DEGs (\u003cb\u003eTable \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003eS2\u003c/span\u003e\u003c/b\u003e). Together, these transcriptomic findings corroborate the metabolomic data, providing convergent evidence that EC-DM tumors are characterized by enhanced carbon metabolism, disrupted redox homeostasis, and altered nucleotide turnover, while also implicating systemic regulatory pathways such as the renin\u0026ndash;angiotensin axis in diabetes-associated endometrial cancer progression.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cb\u003eThe gene\u0026ndash;metabolite interaction network revealed potential molecular mechanisms underlying EC-DM.\u003c/b\u003e \u003c/p\u003e \u003cp\u003eIn the cohort with both metabolomic and transcriptomic data, we performed correlation analyses to investigate the associations between DEMs and DEGs. Significant correlations (p\u0026thinsp;\u0026le;\u0026thinsp;0.01) were observed in pathways related to the monocarboxylic acid metabolic process, terpenoid metabolic process, and the renin\u0026ndash;angiotensin system (\u003cb\u003eFig. S5\u003c/b\u003e). Most of these correlations were positive, indicating coordinated regulation between metabolites and genes involved in cellular respiration and energy metabolism across transcriptomic and metabolomic levels.\u003c/p\u003e \u003cp\u003eTo further integrate metabolomic and transcriptomic datasets, we constructed a gene\u0026ndash;metabolite interaction network based on 1,123 DEGs and 55 core differential metabolites. The final network identified 10 upregulated genes and 7 downregulated genes interacting with 12 differential metabolites (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). These interactions highlight the close functional links between genes and metabolites and provide novel evidence for understanding the molecular mechanisms driving EC-DM progression. The specific functional annotations of these genes are summarized in \u003cb\u003eTable S5\u003c/b\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study represents a comprehensive multi-omics investigation into the metabolic and transcriptomic differences between endometrial cancer patients with and without diabetes. We identified significant alterations in purine metabolism, central carbon metabolism, and immune-related pathways in EC-DM, supported by both metabolite and gene expression data.\u003c/p\u003e \u003cp\u003eA prominent feature of EC-DM tumors was the elevation of glycolytic and TCA intermediates, including pyruvic and citric acid (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). This pattern suggests enhanced glycolytic flux and mitochondrial oxidation, reflecting a hyperglycemia-driven metabolic reprogramming. Chronic hyperglycemia and hyperinsulinemia activate the PI3K/Akt/mTOR axis, which upregulates glycolytic enzymes and promotes the Warburg effect even under normoxic conditions[\u003cspan additionalcitationids=\"CR33\" citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. In endometrial tissue, this metabolic plasticity may enable tumor cells to adapt to oxidative stress and sustain rapid proliferation. The observed upregulation of genes associated with monocarboxylic acid metabolism further supports an intensified carbon flux toward energy production and biosynthesis. These metabolic adjustments, while promoting tumor growth, may also increase dependence on specific pathways\u0026mdash;offering potential metabolic vulnerabilities for therapeutic targeting.\u003c/p\u003e \u003cp\u003eAnother defining alteration in EC-DM was the consistent downregulation of purine degradation intermediates such as uric acid, xanthine, and guanine. Uric acid, a key end-product of purine metabolism, functions as a double-edged molecule with both antioxidant and pro-oxidant roles depending on the cellular context. The reduced levels of purine metabolites observed here likely indicate impaired antioxidant buffering capacity under diabetic conditions (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), where chronic oxidative stress suppresses xanthine oxidase activity and perturbs nucleotide turnover[\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. This imbalance may heighten DNA oxidative damage and genomic instability, contributing to tumor aggressiveness in EC-DM. Xanthine oxidoreductase (XOR) has been shown to affect cell growth in tumors through ROS production, suggesting its role in cancer pathology[\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e], reinforcing its significance as both a biomarker and potential metabolic target.\u003c/p\u003e \u003cp\u003eThe gene-metabolite interaction network highlighted cross-talk between metabolic and immune pathways, with IL6, PPARG, and PHGDH emerging as central players. IL6 is a central cytokine in obesity- and diabetes-associated inflammation, known to activate the JAK/STAT3 pathway and promote angiogenesis and immune evasion[\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. Upregulation of IL6 in EC-DM tumors likely contributes to the establishment of a pro-inflammatory, pro-tumorigenic microenvironment. In parallel, dysregulation of PPARG, a master regulator of lipid metabolism and insulin sensitivity, suggests reprogramming of fatty acid oxidation and lipid biosynthesis in response to altered glucose\u0026ndash;lipid homeostasis. Meanwhile, PHGDH catalyzes the first step in the serine synthesis pathway, providing one-carbon units and NADPH for nucleotide synthesis and redox control[\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. The coordinated activation of IL6, PPARG, and PHGDH thus delineates a metabolic\u0026ndash;immune axis through which diabetes amplifies anabolic metabolism and inflammatory signaling, jointly driving endometrial tumor progression.\u003c/p\u003e \u003cp\u003eThe identification of robust metabolic biomarkers\u0026mdash;including uric acid, D-malic acid, and guanosine diphosphate\u0026mdash;with AUROC values exceeding 0.8 underscores the diagnostic potential of tissue-derived metabolomic signatures. Compared with circulating metabolites, tissue-level profiles more accurately reflect tumor-specific reprogramming under the diabetic milieu. These findings hold promise for developing predictive biomarkers to distinguish EC-DM from EC-NDM or to monitor metabolic therapy response.\u003c/p\u003e \u003cp\u003eThe main limitations of this study include its relatively small sample size and lack of mechanistic validation. However, the use of tumor tissue\u0026ndash;based multi-omics and machine learning provides high specificity and internal consistency, compensating for sample number constraints. Future studies should validate these findings in larger cohorts and use experimental models to confirm the functional roles of key biomarkers.\u003c/p\u003e \u003cp\u003eIn conclusion, our integrated metabolomic and transcriptomic analysis reveals that diabetes significantly alters the metabolic landscape of endometrial cancer, identifying robust biomarkers and dysregulated pathways that may inform future diagnostic and therapeutic strategies.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eOur study demonstrates that endometrial cancer patients with diabetes (EC-DM) exhibit distinct metabolic and transcriptomic profiles compared to those without diabetes (EC-NDM). Key dysregulated pathways include purine metabolism, TCA cycle, amino acid metabolism, and immune response. Machine learning identified robust diagnostic biomarkers such as uric acid, D-malic acid, and guanosine diphosphate. Integrated analysis revealed coordinated changes in genes and metabolites involved in carbon metabolism, redox homeostasis, and the renin-angiotensin system. These findings suggest that diabetes alters the metabolic landscape of endometrial cancer, offering potential biomarkers for diagnosis and insights into therapeutic targeting.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eW Zhang\u003c/strong\u003e: Data curation, Software, Investigation, Writing-original draft, Funding acquisition. \u003cstrong\u003eX Liu\u003c/strong\u003e: Data curation, Software, Investigation, Formal analysis. \u003cstrong\u003eM Fu\u003c/strong\u003e: Methodology, writing\u0026ndash;review and editing.\u003cstrong\u003e\u0026nbsp;S Chen\u003c/strong\u003e: Methodology, Formal analysis, Writing - original draft.\u003cstrong\u003e\u0026nbsp;J Tian\u003c/strong\u003e: Conceptualization, Project administration, Investigation, Resources.\u003cstrong\u003e\u0026nbsp;Y Liu\u003c/strong\u003e:Conceptualization, Project administration, Investigation, Resources. \u003cstrong\u003eP Qu\u003c/strong\u003e: Conceptualization, Project administration, Supervision.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by Tianjin Science and Technology Planning Project (24JCYBJC01650 and 21JCQNJC00230)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe datasets generated and/or analyzed during the current study have been deposited in the OMIX database of the National Genomics Data Center (NGDC), China National Center for Bioinformation, under accession number OMIX013377.\u003c/p\u003e\n\u003cp\u003eDue to regulations concerning human genetic resources, the data are available under controlled access. The metadata of the dataset are publicly accessible at: https://ngdc.cncb.ac.cn/omix/preview/hAVqoNUo.\u003c/p\u003e\n\u003cp\u003eAccess to the full datasets can be granted to qualified researchers through the NGDC data access application process. Full public release of the data will be completed after approval by the relevant regulatory authorities.\u003c/p\u003e\n\u003cp\u003eThe following publicly available databases were used in this study: Kyoto Encyclopedia of Genes and Genomes (https://www.genome.jp/kegg/), Human Metabolome Database (https://hmdb.ca/), Small Molecule Pathway Database ( https://smpdb.ca/).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eOnstad MA, Schmandt RE, Lu KH. Addressing the Role of Obesity in Endometrial Cancer Risk, Prevention, and Treatment. J Clin Oncol. 2016;34(35):4225\u0026ndash;30.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eForte M, Cecere SC, Di Napoli M, Ventriglia J, Tambaro R, Rossetti S, Passarelli A, Casartelli C, Rauso M, Alberico G, et al. Endometrial cancer in the elderly: Characteristics, prognostic and risk factors, and treatment options. Crit Rev Oncol Hematol. 2024;204:104533.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcVicker L, Cardwell CR, Edge L, McCluggage WG, Quinn D, Wylie J, McMenamin \u0026Uacute;C. Survival outcomes in endometrial cancer patients according to diabetes: a systematic review and meta-analysis. BMC Cancer. 2022;22(1):427.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePeeri NC, Bertrand KA, Na R, De Vivo I, Setiawan VW, Seshan VE, Alemany L, Chen Y, Clarke MA, Clendenen T, et al. Understanding risk factors for endometrial cancer in young women. J Natl Cancer Inst. 2025;117(1):76\u0026ndash;88.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFriberg E, Orsini N, Mantzoros CS, Wolk A. Diabetes mellitus and risk of endometrial cancer: a meta-analysis. Diabetologia. 2007;50(7):1365\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang ZH, Su PY, Hao JH, Sun YH. The role of preexisting diabetes mellitus on incidence and mortality of endometrial cancer: a meta-analysis of prospective cohort studies. Int J Gynecol Cancer. 2013;23(2):294\u0026ndash;303.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLuo J, Beresford S, Chen C, Chlebowski R, Garcia L, Kuller L, Regier M, Wactawski-Wende J, Margolis KL. Association between diabetes, diabetes treatment and risk of developing endometrial cancer. Br J Cancer. 2014;111(7):1432\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNief CA, Long SE, McCleary TL, Kidd E, Litkouhi B, Howitt BE. Diabetes mellitus complications associated with recurrence of stage I endometrioid endometrial cancer: A single-center retrospective study. Gynecol Oncol. 2024;190:298\u0026ndash;306.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEsposito K, Chiodini P, Capuano A, Bellastella G, Maiorino MI, Giugliano D. Metabolic syndrome and endometrial cancer: a meta-analysis. Endocrine. 2014;45(1):28\u0026ndash;36.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJin Q, Ma RCW. Metabolomics in Diabetes and Diabetic Complications: Insights from Epidemiological Studies. Cells 2021, 10(11).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGogna N, Krishna M, Oommen AM, Dorai K. Investigating correlations in the altered metabolic profiles of obese and diabetic subjects in a South Indian Asian population using an NMR-based metabolomic approach. Mol Biosyst. 2015;11(2):595\u0026ndash;606.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu F, Tavintharan S, Sum CF, Woon K, Lim SC, Ong CN. Metabolic signature shift in type 2 diabetes mellitus revealed by mass spectrometry-based metabolomics. J Clin Endocrinol Metab. 2013;98(6):E1060\u0026ndash;1065.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGong S, Huang R, Wang M, Lian F, Wang Q, Liao Z, Fan C. Comprehensive analysis of the metabolomics and transcriptomics uncovers the dysregulated network and potential biomarkers of Triple Negative Breast Cancer. J Transl Med. 2024;22(1):1016.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePiell KM, Poulton CC, Stanley CG, Schultz DJ, Klinge CM. Integrated Metabolomics and Transcriptomics Analysis of Anacardic Acid Inhibition of Breast Cancer Cell Viability. Int J Mol Sci 2024, 25(13).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang AMY, Xia YH, Lin JSH, Chu KH, Wang WCK, Ruiter TJJ, Yang JCC, Chen N, Chhuor J, Patil S, et al. Hyperinsulinemia acts via acinar insulin receptors to initiate pancreatic cancer by increasing digestive enzyme production and inflammation. Cell Metab. 2023;35(12):2119\u0026ndash;e21352115.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGallagher EJ, LeRoith D. Hyperinsulinaemia in cancer. Nat Rev Cancer. 2020;20(11):629\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChang SC, Yang WV. Hyperglycemia, tumorigenesis, and chronic inflammation. Crit Rev Oncol Hematol. 2016;108:146\u0026ndash;53.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJeong HS, Lee DH, Kim SH, Lee CH, Shin HM, Kim HR, Cho CH. Hyperglycemia-induced oxidative stress promotes tumor metastasis by upregulating vWF expression in endothelial cells through the transcription factor GATA1. Oncogene. 2022;41(11):1634\u0026ndash;46.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLior C, Barki D, Halperin C, Iacobuzio-Donahue CA, Kelsen D, Shouval RS. Mapping the tumor stress network reveals dynamic shifts in the stromal oxidative stress response. Cell Rep. 2024;43(5):114236.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQiu C, Tang C, Tang Y, Su K, Chai X, Zhan Z, Niu X, Li J. RGS5(+) lymphatic endothelial cells facilitate metastasis and acquired drug resistance of breast cancer through oxidative stress-sensing mechanism. Drug Resist Updat. 2024;77:101149.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChakraborty S, Balan M, Sabarwal A, Choueiri TK, Pal S. Metabolic reprogramming in renal cancer: Events of a metabolic disease. Biochim Biophys Acta Rev Cancer. 2021;1876(1):188559.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAsiri A, Al Qarni A, Bakillah A. The Interlinking Metabolic Association between Type 2 Diabetes Mellitus and Cancer: Molecular Mechanisms and Therapeutic Insights. \u003cem\u003eDiagnostics (Basel)\u003c/em\u003e 2024, 14(19).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang X, Li Y, Hou X, Li J, Ma X. Lipid metabolism reprogramming in endometrial cancer: biological functions and therapeutic implications. Cell Commun Signal. 2024;22(1):436.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRazghonova Y, Mika A, Czapiewska M, Stanczak A, Zygowska P, Wydra DG, Sledzinski T, Abacjew-Chmylko A. Endometrial Cancer Is Associated with Altered Metabolism and Composition of Fatty Acids. Int J Mol Sci 2025, 26(7).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKliemann N, Viallon V, Murphy N, Beeken RJ, Rothwell JA, Rinaldi S, Assi N, van Roekel EH, Schmidt JA, Borch KB, et al. Metabolic signatures of greater body size and their associations with risk of colorectal and endometrial cancers in the European Prospective Investigation into Cancer and Nutrition. BMC Med. 2021;19(1):101.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDossus L, Kouloura E, Biessy C, Viallon V, Siskos AP, Dimou N, Rinaldi S, Merritt MA, Allen N, Fortner R, et al. Prospective analysis of circulating metabolites and endometrial cancer risk. Gynecol Oncol. 2021;162(2):475\u0026ndash;81.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSidorkiewicz I, J\u0026oacute;źwik M, Buczyńska A, Erol A, J\u0026oacute;źwik M, Moniuszko M, Jarząbek K, Niemira M, Krętowski A. Identification and subsequent validation of transcriptomic signature associated with metabolic status in endometrial cancer. Sci Rep. 2023;13(1):13763.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFan X, Zou X, Liu C, Cheng W, Zhang S, Geng X, Zhu W. MicroRNA expression profile in serum reveals novel diagnostic biomarkers for endometrial cancer. Biosci Rep 2021, 41(6).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKnific T, Vouk K, Smrkolj Š, Prehn C, Adamski J, Rižner TL. Models including plasma levels of sphingomyelins and phosphatidylcholines as diagnostic and prognostic biomarkers of endometrial cancer. J Steroid Biochem Mol Biol. 2018;178:312\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYan X, Zhao W, Wei J, Yao Y, Sun G, Wang L, Zhang W, Chen S, Zhou W, Zhao H, et al. A serum lipidomics study for the identification of specific biomarkers for endometrial polyps to distinguish them from endometrial cancer or hyperplasia. Int J Cancer. 2022;150(9):1549\u0026ndash;59.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTang YT, Huang YY, Zheng L, Qin SH, Xu XP, An TX, Xu Y, Wu YS, Hu XM, Ping BH, et al. Comparison of isolation methods of exosomes and exosomal RNA from cell culture medium and serum. Int J Mol Med. 2017;40(3):834\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDeBerardinis RJ, Chandel NS. Fundamentals of cancer metabolism. Sci Adv. 2016;2(5):e1600200.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu S, Zhang H, Gao C, Chen J, Li H, Meng Z, Bai J, Shen Q, Wu H, Yin T. Hyperglycemia Enhances Immunosuppression and Aerobic Glycolysis of Pancreatic Cancer Through Upregulating Bmi1-UPF1-HK2 Pathway. Cell Mol Gastroenterol Hepatol. 2022;14(5):1146\u0026ndash;65.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSu Y, Luo Y, Zhang P, Lin H, Pu W, Zhang H, Wang H, Hao Y, Xiao Y, Zhang X, et al. Glucose-induced CRL4(COP1)-p53 axis amplifies glycometabolism to drive tumorigenesis. Mol Cell. 2023;83(13):2316\u0026ndash;e23312317.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSautin YY, Johnson RJ. Uric acid: the oxidant-antioxidant paradox. Nucleosides Nucleotides Nucleic Acids. 2008;27(6):608\u0026ndash;19.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYan S, Zhang P, Xu W, Liu Y, Wang B, Jiang T, Hua C, Wang X, Xu D, Sun B. Serum Uric Acid Increases Risk of Cancer Incidence and Mortality: A Systematic Review and Meta-Analysis. \u003cem\u003eMediators Inflamm\u003c/em\u003e 2015, 2015:764250.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAgarwal A, Banerjee A, Banerjee UC. Xanthine oxidoreductase: a journey from purine metabolism to cardiovascular excitation-contraction coupling. Crit Rev Biotechnol. 2011;31(3):264\u0026ndash;80.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChan LC, Li CW, Xia W, Hsu JM, Lee HH, Cha JH, Wang HL, Yang WH, Yen EY, Chang WC, et al. IL-6/JAK1 pathway drives PD-L1 Y112 phosphorylation to promote cancer immune evasion. J Clin Invest. 2019;129(8):3324\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCatar R, Witowski J, Zhu N, L\u0026uuml;cht C, Derrac Soria A, Uceda Fernandez J, Chen L, Jones SA, Fielding CA, Rudolf A, et al. IL-6 Trans-Signaling Links Inflammation with Angiogenesis in the Peritoneal Membrane. J Am Soc Nephrol. 2017;28(4):1188\u0026ndash;99.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAhmadian M, Suh JM, Hah N, Liddle C, Atkins AR, Downes M, Evans RM. PPARγ signaling and metabolism: the good, the bad and the future. Nat Med. 2013;19(5):557\u0026ndash;66.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8035766/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8035766/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eEndometrial cancer (EC) patients with type 2 diabetes mellitus (DM) often exhibit a more aggressive tumor phenotype and poorer prognosis. However, the underlying metabolic and molecular mechanisms remain poorly understood. Objective: This study aimed to identify diagnostic biomarkers and dysregulated pathways in EC patients with diabetes (EC-DM) through integrated metabolomic and transcriptomic analyses of tumor tissues.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eTumor tissues from 20 EC patients (10 EC-DM, 10 non-DM) were analyzed. Untargeted metabolomics used LC\u0026ndash;HRMS, and transcriptomics used RNA-seq in 12 patients. Differentially expressed metabolites and genes were identified via machine learning and DESeq2. Pathway and multi-omics integration were performed.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eMultivariate analysis revealed distinct metabolic profiles between EC-DM and EC-NDM groups. Machine learning identified robust DEMs including uric acid, D-malic acid, and guanosine 5\u0026prime;-diphosphate. Pathway analysis showed significant enrichment in purine metabolism, citrate cycle, arginine biosynthesis, and glycerophospholipid metabolism in EC-DM. Transcriptomics identified 1,123 DEGs, with enrichment in monocarboxylic acid metabolic process, terpenoid metabolism, and renin\u0026ndash;angiotensin system. Integrated gene\u0026ndash;metabolite interaction networks revealed key interactions involving \u0026ldquo;IL6\u0026rdquo;, \u0026ldquo;PPARg\u0026rdquo;, \u0026ldquo;PHGDH\u0026rdquo;, and purine metabolites.\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eOur study demonstrates that diabetes reprograms tumor metabolism in endometrial cancer, leading to distinct metabolic and transcriptional alterations. We identified potential diagnostic biomarkers and highlighted dysregulated pathways that may underlie the aggressive phenotype of EC-DM. These findings provide insights into the metabolic interplay between diabetes and endometrial cancer and offer candidates for further validation.\u003c/p\u003e","manuscriptTitle":"Integrated tumor tissue metabolomic and transcriptomic approaches for identifying diagnostic biomarkers in endometrial cancer with diabetes mellitus","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-16 08:22:01","doi":"10.21203/rs.3.rs-8035766/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"4cca9aa9-82a2-4dfb-af47-8f05769d3075","owner":[],"postedDate":"January 16th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-03-04T11:41:58+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-16 08:22:01","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8035766","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8035766","identity":"rs-8035766","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}