Linking metabolism and metastasis: elevated α-hydroxybutyric acid in oral squamous cell carcinoma patients with lymph node metastasis - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Metabolomics . 2026 Apr 18;22(3):55. doi: 10.1007/s11306-026-02431-7 Search in PMC Search in PubMed View in NLM Catalog Add to search Linking metabolism and metastasis: elevated α-hydroxybutyric acid in oral squamous cell carcinoma patients with lymph node metastasis Xiaolian Gu Xiaolian Gu 1 Department of Medical Biosciences/Pathology, Umeå University, Building 6M, 2nd floor, Analysvägen 9, Umeå, 901 87 Västerbotten Sweden Find articles by Xiaolian Gu 1, ✉ , Philip J Coates Philip J Coates 2 Research Centre for Applied Molecular Oncology (RECAMO), Masaryk Memorial Cancer Institute, Brno, 656 53 Czech Republic Find articles by Philip J Coates 2 , Lixiao Wang Lixiao Wang 1 Department of Medical Biosciences/Pathology, Umeå University, Building 6M, 2nd floor, Analysvägen 9, Umeå, 901 87 Västerbotten Sweden Find articles by Lixiao Wang 1 , Nicola Sgaramella Nicola Sgaramella 1 Department of Medical Biosciences/Pathology, Umeå University, Building 6M, 2nd floor, Analysvägen 9, Umeå, 901 87 Västerbotten Sweden 3 Department of Oral and Maxillo-Facial Surgery, Mater Dei Hospital, Bari, 701 25 Italy Find articles by Nicola Sgaramella 1, 3 , Mustafa Magan Mustafa Magan 1 Department of Medical Biosciences/Pathology, Umeå University, Building 6M, 2nd floor, Analysvägen 9, Umeå, 901 87 Västerbotten Sweden 4 Department of Clinical Sciences/ENT, Umeå University, Umeå, 901 87 Västerbotten Sweden Find articles by Mustafa Magan 1, 4 , Karin Nylander Karin Nylander 1 Department of Medical Biosciences/Pathology, Umeå University, Building 6M, 2nd floor, Analysvägen 9, Umeå, 901 87 Västerbotten Sweden Find articles by Karin Nylander 1 Author information Article notes Copyright and License information 1 Department of Medical Biosciences/Pathology, Umeå University, Building 6M, 2nd floor, Analysvägen 9, Umeå, 901 87 Västerbotten Sweden 2 Research Centre for Applied Molecular Oncology (RECAMO), Masaryk Memorial Cancer Institute, Brno, 656 53 Czech Republic 3 Department of Oral and Maxillo-Facial Surgery, Mater Dei Hospital, Bari, 701 25 Italy 4 Department of Clinical Sciences/ENT, Umeå University, Umeå, 901 87 Västerbotten Sweden ✉ Corresponding author. Received 2025 Jul 28; Accepted 2026 Apr 8; Issue date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . PMC Copyright notice PMCID: PMC13091896 PMID: 41999538 Abstract Introduction Metabolic reprogramming is a hallmark of cancer. Plasma metabolomics offers a minimally invasive approach for identifying metabolic alterations that may provide insights into tumor progression. Objectives We aimed to characterize plasma metabolomic profiles in patients with oral squamous cell carcinoma (OSCC) and evaluate their clinical relevance. Methods Plasma samples from 43 OSCC patients and 129 cancer-free controls, matched at a 1:3 ratio based on age, sex, and body mass index, were analyzed using gas chromatography-mass spectrometry (GC-MS). A random forest algorithm was applied to identify key metabolic features distinguishing OSCC from controls. The clinical significance of the top metabolites was assessed and validated in another OSCC cohort ( n = 27). Results A total of 113 compounds were putatively annotated and analyzed based on relative abundances. A ten-feature panel demonstrated good classification performance (area under the curve = 0.87; Matthews correlation coefficient = 0.703). The ten features are maltose, glucose, xylulose, δ-gluconolactone, fructose, indoleacetic acid, α-hydroxybutyric acid, glutamic acid, cysteine, and the monoacylglyceride MG(18:1(9Z)/0:0/0:0), suggesting dysregulated carbohydrate metabolism and oxidative stress as the major plasma metabolomic alterations in OSCC. Notably, α-hydroxybutyric acid levels were elevated in patients with regional lymph node metastasis compared with those without. Conclusion Our findings underscore the intricate interplay between altered glucose metabolism, redox imbalance, and OSCC. α-hydroxybutyric acid, a marker of oxidative stress and an indicator of insulin resistance, may be associated with metastatic progression. Supplementary Information The online version contains supplementary material available at 10.1007/s11306-026-02431-7. Keywords: Oral cancer, Plasma, Metabolomics, Glucose, α-hydroxybutyric acid Background Metabolomics, the high-throughput study of small-molecule metabolites in biological systems, has emerged as a powerful tool for uncovering the biochemical alterations associated with disease (Danzi et al., 2023 ; Wishart, 2016 ). Since metabolites represent the intermediates and end-products of cellular processes, metabolomic profiling provides a snapshot of an organism’s physiological state, reflecting both intrinsic and extrinsic influences such as genetic/epigenetic alterations, nutrients, drugs, lifestyle, and environmental conditions (Gupta et al., 2024 ; Schmidt et al., 2021 ). Metabolic reprogramming is a hallmark feature of cancer, with tumors often exhibiting alterations in pathways related to energy production, lipid metabolism, and amino acid biosynthesis (Gupta et al., 2024 ; Schmidt et al., 2021 ). These metabolic changes can be detected in biofluids such as plasma, making metabolomics a promising approach for non-invasive cancer biomarker discovery (Chen et al., 2024 ; Irajizad et al., 2023 ; Wang et al., 2023 ). Oral squamous cell carcinoma (OSCC) is the most common type of oral cancer, arising from squamous epithelial cells lining the oral cavity (Tan et al., 2023 ). While metabolomic studies have provided insights into the biochemical alterations in various cancers, the metabolic profile of patients with OSCC remains less well-characterized. Saliva, serum, plasma, urine, and tissue samples from OSCC patients have been investigated, with findings showing considerable variability regarding the specific metabolites and metabolic pathways implicated in oral carcinogenesis, due to differences in study design, analytical platforms, and/or data processing methods (Alapati et al., 2023 ; Luo et al., 2024 ; Vitorio et al., 2020 ). This lack of consistency underscores the need for further research to identify robust and reproducible metabolic biomarkers for OSCC. Among the various analytical platforms available, gas chromatography-mass spectrometry (GC-MS)-based metabolomic technologies are widely employed due to their high sensitivity, resolution, and ability to detect a wide range of volatile and semi-volatile metabolites. With appropriate derivatization, they can also detect metabolites that are inherently non-volatile, thereby enabling the reliable analysis of a broad spectrum of metabolites involved in central metabolic pathways (Danzi et al., 2023 ). In this study, targeted GC-MS was employed to profile plasma metabolites in patients with OSCC, aiming to uncover underlying metabolic alterations that could support the development of novel biomarkers for early detection and disease monitoring. Materials and methods OSCC patients (1st patient cohort) and matched controls Forty-three patients diagnosed with OSCC at Norrland’s University Hospital in Umeå, Sweden were recruited. Blood samples were collected in heparin tubes prior to treatment. Plasma was aliquoted and stored at -80 °C within 1 h of collection. A total of 129 cancer-free individuals were matched to the 43 patients at a 3:1 ratio based on age, sex, and body mass index (BMI). Heparin plasma samples from the cancer-free individuals were obtained from Biobank Norr, Umeå. For these samples, blood collection and processing were performed following a standardized protocol. Briefly, after a minimum of 8 h of fasting, blood samples were collected in heparin tubes in the morning. Plasma samples were aliquoted and frozen within 1 h of collection, either directly at -80 °C or at -20 °C for up to one week before being transferred to a central storage facility. OSCC patients without matched controls (2nd patient cohort) An additional cohort of 27 OSCC patients was recruited. No cancer-free controls were matched to these patients. Blood samples were collected in heparin tubes prior to treatment. Demographic and clinical features of both patient cohorts are presented in Table 1 . Except for differences in age and fasting status, the characteristics of patients in the two cohorts were comparable. Table 1. Demographic and clinical features of controls and the two patient cohorts Control Patient p -value (Fisher’s exact test) 1st cohort (control-matched) 2nd cohort (non-matched) Control vs. 1st cohort 1st vs. 2nd cohort Age ≤ 40 years 39 13 0 1 < 0.001 > 40 years 86 29 27 Sex Female 44 15 14 1 0.218 Male 81 27 13 Obesity No 88 30 22 1 0.403 Yes 37 12 5 Hypertension No 71 31 22 0.067 0.565 Yes 54 11 5 Diabetes * No 108 36 25 0.070 0.467 Yes 5 6 2 Fasting status Fasting 125 31 26 <0.001 0.021 Non-fasting 0 11 1 Tumor subsite Oral tongue 29 13 0.129 Other oral subsites 13 14 Tumor size T1, T2 28 13 0.141 T3, T4 14 14 lymph node metastasis Negative 27 18 1.000 Positive 15 9 TNM stage I, II 21 11 0.471 III, IV 21 16 Open in a new tab OSCC oral squamous cell carcinoma *Diabetes status was unavailable for 12 controls Ethics This study was approved by the Regional Ethics Review Board, Umeå, Sweden (Dnr 08-003 M, 2012-131-33 M and 2024-00725-02) and conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants. Metabolomics analysis Plasma samples were sent to The Swedish Metabolomics Centre in Umeå, Sweden ( www.swedishmetabolomicscentre.se ) for metabolic profiling using Gas Chromatography - Time-of-Flight - Mass Spectrometry (GC-TOF-MS). Each set of case and matched control samples was measured consecutively to minimize measurement order bias. Non-matched case samples were measured after all case-control sets. Metabolite annotation and quantification were performed jointly for samples from both cohorts. Internal standards alpha-ketoglutarate-13C4, myristic acid-13C3, cholesterol-D7 were obtained from Cil (Andover, MA, USA). Succinic acid-D4, hexadecanoic acid-13C4, D-glucose-13C6, D-sucrose-13C12 were obtained from Sigma (St. Louis, MO, USA). Metabolites were extracted by mixing 80 µL plasma with 720 µL 90:10 (v/v) methanol: water solution containing internal standards. Each plate was extracted for 30 s using a mixing script on an Agilent Bravo automated liquid handling platform, incubated in the freezer at -20 °C for 2 h and centrifuged at 4 °C 14,000 RPM, for 10 min. Derivatization and GC-MS analysis were performed as described previously (Gullberg et al., 2004 ) with the samples being derivatized in 15 µl methoxyamine (15 µg/µl in pyridine) and 30 µl of a 1:1 mix of MSTFA (+ 1% TMCS) and heptane including methyl stearate (15 ng/µl). 1 µL of the derivatized sample was injected in splitless mode by a L-PAL3 autosampler (CTC Analytics AG, Switzerland) into an Agilent 7890B gas chromatograph equipped with a 10 m x 0.18 mm fused silica capillary column with a chemically bonded 0.18 μm Rxi-5 Sil MS stationary phase (Restek Corporation, U.S.) The injector temperature was 270 °C, the purge flow rate was 20 mL min-1 and the purge was turned on after 60 s. The gas flow rate through the column was 1 mL min-1, the column temperature was held at 70 °C for 2 min, then increased by 40 °C min-1 to 320 °C, and held there for 2 min. The column effluent was introduced into the ion source of a Pegasus BT time-of-flight mass spectrometer, GC-TOF-MS (Leco Corp., St Joseph, MI, USA). The transfer line and the ion source temperatures were 250 °C and 200 °C, respectively. Ions were generated by a 70 eV electron beam at an ionization current of 2.0 mA, and 30 spectra s-1 were recorded in the mass range m/z 50–800. The acceleration voltage was turned on after a solvent delay of 150 s. The detector voltage was 1800–2300 V. All non-processed MS-files were exported from the ChromaTOF software in NetCDF format to MATLAB R2023a (Mathworks, Natick, MA, USA), where all data pre-treatment procedures, such as base-line correction, chromatogram alignment, data compression and Multivariate Curve Resolution were performed. The extracted mass spectra were identified by comparisons of their retention index and mass spectra with libraries of retention time indices and mass spectra (Schauer et al., 2005 ) using NIST MS 2.2 software. Compound annotation was based on forward and reverse spectral matching against a curated in-house GC-MS metabolite library developed at the Swedish Metabolomics Centre, Umeå University. Masses and ratio between masses indicative of a derivatized metabolite were especially notified. The mass spectrum with the highest probability indicative of a metabolite and the retention index between the sample and library for the suggested metabolite was ± 5 (usually less than 3) the deconvoluted “peak” was annotated as an identification of a metabolite. Metabolite annotations therefore correspond to Metabolomics Standards Initiative (MSI) level 2 (putatively annotated compounds) and were manually reviewed by experienced analysts at the Swedish Metabolomics Centre. Chromatographic peaks were resolved using spectral deconvolution, and for each metabolite a compound-specific fragment ion was selected for quantification. Metabolite abundances were calculated as the integrated peak area of the selected fragment ion. Internal standards (IS02 and IS07; Table S1) were added prior to extraction and derivatization to monitor analytical performance. Seven pooled quality control (QC) samples were included in the analytical batch to assess analytical stability and reproducibility during the measurement sequence. As the analysis provides relative quantification, metabolite abundances were scaled so that the maximum value for each metabolite across samples was set to 100 to facilitate comparison of relative changes among samples. Data analysis Metabolomic data were analyzed using MetaboAnalyst, a web-based platform for comprehensive metabolomics data analysis (Ewald et al., 2024 ; Pang et al., 2024 ). Original relative abundance data from the first cohort were uploaded and normalized using log transformation (base 10) and auto-scaling. To identify differences between patients and matched controls, univariate analysis was performed using t-tests, with p -values adjusted for multiple comparisons using the Benjamini–Hochberg false discovery rate (FDR) method. Principal component analysis (PCA) was used for linear dimensionality reduction, and the Random Forest algorithm was applied to capture potential non-linear relationships (Greener et al., 2022 ). All analyses were performed using the normalized data and default parameters in MetaboAnalyst. Model performance was evaluated using cross-validation, as implemented in MetaboAnalyst. In addition, model stability was assessed using the out-of-bag (OOB) error estimate inherent to the Random Forest algorithm, which provides internal validation based on bootstrap resampling. For analyses relating clinical features to metabolomic alterations, original relative abundance data were used. Associations between two binary variables were assessed using Fisher’s exact test. To compare numerical variables between two groups of samples, the non-parametric Mann-Whitney U test was used. Survival analysis was performed using the Kaplan‑Meier method and Cox-regression. Two survival measures were assessed: overall survival (time from diagnosis to death from any cause) and disease‑free survival (time from completion of treatment to relapse or death). Statistical tests were conducted in IBM SPSS Statistics 28 (IBM Corp., Armonk, NY, USA). A two-sided p -value < 0.05 was considered statistically significant. Box plots were created using ggplot2 (R version 4.5.0). Results GC-TOF-MS data A total of 199 samples were processed for metabolomics profiling. However, samples from one patient in the first cohort and one control could not be analyzed. To maintain the intended case–control matching, the three control samples matched to the failed patient sample were excluded. As a result, 194 samples were included in the final data analysis: 42 patients from the first cohort, 125 matched controls, and 27 patients from the second cohort (Table S1). PCA analysis of all samples and seven QC samples indicated no major analytical issues, although the first five QC samples clustered closely while the last two were more distant, reflecting a measurement order bias (Fig. S1 ). The non-matched cases from the second patient cohort were measured after all case-control sets from the first cohort; therefore, the two patient cohorts were not compared directly. In total, 113 compounds were annotated, including organic acids, lipids, and organic oxygen compounds (Fig. S2 ). Drug-related metabolites Statistically significant differences in 38 compounds were identified between 42 patients and the 125 matched controls using t-test (adjusted p -value < 0.05, Table 2 ). Notably, patients exhibited higher plasma levels of acetaminophen compared to controls. Acetaminophen is a synthetic drug commonly used for pain relief and fever reduction, and we identified that a further seven compounds were highly correlated with acetaminophen ( r > 0.5). These compounds correspond to the top eight differentially abundant metabolites between patients and controls. Among these, p -cresol glucuronide is a conjugated metabolite of p -cresol, sharing primary metabolic pathways with acetaminophen. Capric acid is used in various nutritional supplements and medical formulations, and tricaprylin, glyceryl 2-caprate dicaprylate, and glyceryl 2-caprylate dicaprate are lipid-based excipients commonly employed in drug formulations to enhance solubility and bioavailability of poorly water-soluble drugs. Table 2. Differentially abundant compounds between patients and controls Compound Fold change (patient/control) Adjusted p -value p-Cresol glucuronide 2872.40 5.36E-22 Acetaminophen 12.14 5.36E-22 Isomaltose 71.46 3.93E-17 Capric acid 7.84 1.37E-13 Tricaprylin 154.04 3.17E-12 Glyceryl 2-caprylate dicaprate 105.94 3.19E-12 Glyceryl 2-caprate dicaprylate 200.13 3.56E-12 δ-tocopherol 3.98 5.66E-08 Maltose 1.95 7.25E-08 Xylulose 1.37 3.50E-07 Glucose 1.13 3.97E-07 Fructose 2.62 6.73E-06 α-hydroxybutyric acid (2-HBA) 1.57 6.90E-06 Indoleacetic acid (IAA) 0.68 1.03E-05 δ-gluconolactone 2.07 2.69E-05 MG(18:1(9Z)/0:0/0:0) 1.71 2.69E-05 Glutamic acid 1.55 9.89E-05 Sorbitol 8.77 5.27E-04 Cysteine 0.87 6.70E-04 Maltotriose 1.62 7.50E-04 Xylitol 3.31 8.19E-04 Threonic acid 1.29 8.19E-04 α-Tocopherol 0.82 8.91E-04 Glycerol 3-phosphate 1.18 1.26E-03 Hypoxanthine 1.43 1.82E-03 Erythritol 1.31 1.93E-03 Lactic acid 1.16 1.93E-03 α-hydroxyisovaleric acid 1.47 2.28E-03 MG(16:0/0:0/0:0) 1.06 5.50E-03 Hippuric acid 0.62 5.59E-03 Arginine 0.79 8.36E-03 Oxoglutaric acid 1.23 1.14E-02 1,5-anhydrosorbitol (1,5-AG) 0.89 1.51E-02 Niacinamide 1.38 1.90E-02 Glucaric acid 16.23 1.94E-02 β-hydroxybutyric acid (3-HBA) 1.82 2.70E-02 Citric acid 0.89 2.84E-02 Benzoic acid 1.09 4.98E-02 Open in a new tab Traces of ibuprofen, widely used for pain and fever, and metformin, prescribed for type 2 and gestational diabetes, were detected in certain individuals. However, no significant differences in their levels were observed between patients and controls. Since the use of Simvastatin ( n = 4), Levaxin (Levothyroxine, n = 6), and Metformin ( n = 4) was known in a few patients, we further investigated the correlation between intake of these drugs and plasma levels of significantly altered metabolites. No significant associations were identified (data not shown). To identify endogenous cancer-associated metabolomic changes, all drug or drug-related compounds were excluded from subsequent analysis. Based on the remaining 103 metabolites, a volcano plot illustrating the overall differences in metabolite abundance between patients and matched controls is shown in Fig. S3. Principal component analysis (PCA) PCA based on the 103 metabolomic features revealed similar profiles between the 42 patients and the matched 125 controls (Fig. S4). Random forest modeling We explored metabolomic changes in the 42 patients using the random forest classification method. Metabolites were ranked according to their importance scores, and the top N metabolites ( N = 5, 10, 15, 25, 50, or 100) were sequentially selected to evaluate model performance across different feature set sizes. For each feature set, ROC (receiver operating characteristic) curves were plotted to illustrate model performance. The AUC (area under the curve) values of the six models ranged from 0.833 to 0.900 (Fig. 1 a). We further evaluated the performance of the six models using the Matthews correlation coefficient (MCC), a metric particularly well-suited for imbalanced datasets. MCC considers all four outcomes of the confusion matrix: true positives, false positives, true negatives, and false negatives (Chicco & Jurman, 2020 ). Although the 50-feature model showed slightly higher performance for some metrics, the 10-feature model achieved the highest MCC value of 0.703 (Table S2), indicating the most balanced classification performance between the two classes. In addition, the smaller feature set improves model interpretability and reduces the risk of overfitting. Therefore, the 10-feature model was selected as the final model. Fig. 1. Open in a new tab Random forest modeling. a ROC curves and their AUC values based on cross-validation performance showing the performance of six models using 5 to 100 metabolomic features. b Top 10 significant features ranked by average importance measure. ROC receiver-operating characteristic; AUC area under the curve. OSCC oral squamous cell carcinoma. Con matched cancer-free controls KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analysis based on the top 10 important features (Fig. 1 b) identified “starch and sucrose metabolism” as the only significantly enriched pathway (FDR = 0.008). Three metabolites (maltose, glucose, and fructose) were involved in this pathway (Fig. S5). Sub-class chemical structure enrichment analysis revealed significant enrichment of “Carbohydrates and carbohydrate conjugates”, which included maltose, glucose, xylulose, δ-gluconolactone and fructose. All five metabolites exhibited increased plasma levels in patients compared to controls. Other metabolites among the top ten included two organic acids (indoleacetic acid and alpha-hydroxybutyric acid), two amino acids (glutamic acid and cysteine) and one lipid molecule (MG(18:1(9Z)/0:0/0:0)). α-hydroxybutyric acid, glutamic acid and MG(18:1(9Z)/0:0/0:0) showed increased levels, whereas indoleacetic acid and cysteine showed decreased levels in patients compared to controls. The paired sample box plots illustrate consistent patterns of changes in most case-control pairs (Fig. 2 ). Fig. 2. Open in a new tab Paired sample box plots showing plasma levels of the top 10 important metabolic features in patients with oral squamous cell carcinoma (OSCC) and matched controls. Adj. P: adjusted p-value Clinical features and metabolomic alterations The top ten metabolites were analyzed for differences in abundance related to age (young vs. old), sex (female vs. male), fasting status (fasted vs. non-fasted), tumor subsite (oral tongue vs. other oral subsites), tumor size (T1/T2 vs. T3/T4), and nodal status (negative vs. positive) in the first patient cohort, with findings validated in the second cohort. The two patient cohorts were analyzed separately to account for potential batch effects related to measurement order. Given the limited sample sizes, we adopted a conservative approach, focusing on metabolites that exhibited consistent trends across cohorts. Overall, Mann-Whitney U tests revealed no significant differences in metabolite levels based on age, sex, fasting status, tumor subsite or tumor size (Table S3). However, for α-hydroxybutyric acid, levels tended to be higher in patients with lymph node metastasis compared with those without in the first cohort ( p = 0.061, effect size = 0.35, 95% confidence interval = 0.000025–0.628) and confirmed as significantly elevated in the second cohort ( p = 0.046, effect size = 0.48, 95% confidence interval = 0.054–0.76, Fig. 3 ). None of the metabolites demonstrated a significant impact on overall or disease-free survival using Kaplan-Meier or Cox regression survival analysis. Fig. 3. Open in a new tab Box plots showing plasma levels of α-hydroxybutyric acid in node-negative and node-positive patients. a 1st patient cohort. b 2nd patient cohort. Node-negative: patients without regional lymph node metastasis; Node-positive: patients with regional lymph node metastasis Discussion Blood samples were collected at the time of diagnosis, before any cancer treatment had begun. However, we found that levels of drugs and drug-related compounds were significantly higher in patients compared to controls, such as acetaminophen and p-Cresol glucuronide. These increases did not correlate with the reported use of Simvastatin, Levaxin (Levothyroxine) or Metformin by a few patients, which are typically used to lower cholesterol, treat hypothyroidism and control glucose levels, respectively. This suggests that some patients may have engaged in unreported self-medication, possibly to manage pain or improve their health condition. It has previously been shown that drug metabolites detected in plasma frequently do not align with participants’ self-reported medication use (Suhre et al., 2022 ). Therefore, unreported self-medication could represent a confounding factor in metabolomics studies. To focus on endogenous cancer-associated metabolomic changes, drug and drug-related compounds were thus excluded from further analysis. After removing drug-related compounds, our ten-feature random forest model of endogenous metabolites showed good performance in discriminating 42 OSCC from 125 matched controls. Five of the ten features were “carbohydrates and carbohydrate conjugates”, including maltose, glucose, xylulose, δ-gluconolactone and fructose. A common phenomenon in cancer is the Warburg effect, where tumor cells increase their rate of glucose uptake and rely on glycolysis for energy production even in the presence of sufficient oxygen. This metabolic shift results in lower extracellular glucose levels and higher lactate concentrations within the tumor microenvironment (Fendt, 2024 ; Finley, 2023 ; Hardie, 2022 ). Studies analyzing tissue, serum, or saliva samples have reported significantly decreased glucose levels in OSCC patients compared to healthy controls (Enomoto et al., 2018 ; Ogawa et al., 2014 ; Song et al., 2020 ; Tsai et al., 2020 ). However, in accordance with our findings, an increase in serum glucose levels in patients compared to controls has also been reported (Tiziani et al., 2009 ; Yang et al., 2022 ), and elevated plasma glucose was seen in association with cancer development in a rat oral cancer model (Kong et al., 2015 ). These observations align with evidence suggesting a complex interplay between cancer progression and systemic metabolic alterations (Swanton et al., 2024 ). While tumor cells exhibit increased glucose uptake locally, the altered metabolic environment can influence overall glucose homeostasis. One possible explanation for the increased plasma glucose levels is the presence of cancer-associated insulin resistance (Dev et al., 2018 ; Marmol et al., 2023 ; Masi & Patel, 2021 ). This phenomenon may be further compounded by stress-induced hyperglycemia (Dungan et al., 2009 ; Kanter et al., 2024 ). Maltose is a disaccharide composed of two glucose molecules. Xylulose and δ-gluconolactone are intermediates in the Pentose Phosphate Pathway (PPP), which supports nucleotide synthesis and redox balance in cancer cells (Patra & Hay, 2014 ). Fructose is a monosaccharide that is often formed by cleavage of sucrose into fructose and glucose. Overall, increased plasma levels of several common dietary sugars (glucose, fructose, and maltose) and sugar derivatives involved in specific metabolic reactions (xylulose and δ-gluconolactone) indicate that alterations in the metabolism of simple carbohydrates represent the major systemic change in OSCC metabolites. Alterations in three metabolites related to glutathione synthesis, a critical pathway for cellular antioxidant defense and redox homeostasis (Lapenna, 2023 ), were also uncovered in patients: α-hydroxybutyric acid, glutamic acid, and cysteine. α-hydroxybutyric acid and glutamic acid levels were elevated while cysteine levels were decreased in patients. Glutamic acid, known as the primary nutrient source to support tumor growth, is among the most frequently dysregulated metabolites in OSCC patients (Alapati et al., 2023 ). α-hydroxybutyric acid, a byproduct of cysteine synthesis during glutathione production, also serves as a marker of oxidative stress (Lord & Bralley, 2008 ). Moreover, α-hydroxybutyric acid has been established as an early biomarker of insulin resistance and impaired glucose regulation (Gall et al., 2010 ; Sousa et al., 2021 ), and its elevated plasma levels have been reported in certain cancers (Tiziani et al., 2009 ; Zeng et al., 2014 ). Notably, in vivo studies show that α-hydroxybutyric acid not only promotes insulin resistance in a type 2 diabetes mouse model but also exacerbates tumor development in a colorectal cancer mouse model, suggesting that elevated α-hydroxybutyric acid may be a shared metabolic risk factor linking cancer and diabetes (Lv et al., 2023 ). Here, we observed higher α-hydroxybutyric acid levels in node-positive OSCC patients compared to node-negative patients. As lymph node metastasis is commonly associated with increased oxidative stress and enhanced antioxidant capacity (Tasdogan et al., 2021 ), elevated α-hydroxybutyric acid may reflect the metabolic reprogramming that facilitates tumor invasion and lymphatic dissemination. Nevertheless, given the exploratory nature of the analysis and the modest effect size with wide confidence intervals, our findings should be interpreted cautiously and do not establish a causal relationship between α-hydroxybutyric acid and metastatic potential. Given the close interplay between glucose dysregulation, oxidative stress and cancer progression (Giri et al., 2018 ), further studies in larger cohorts and functional models are required to clarify whether α-hydroxybutyric acid plays a mechanistic role in OSCC progression or represents a metabolic marker associated with disease severity. A previous report showed that OSCC patients tend to have a low BMI, with the major tumor subsite being the floor of mouth, followed by the oral tongue (Polachini et al., 2023 ). This was, however, not seen in our patient cohort where the oral tongue was the predominant tumor subsite and SCCOT patients showed a high BMI with one-third being obese. Differences in the distribution of tumor subsites across studies may explain the discrepancies and suggest tumor subsite-specific metabolic alterations. Alternative metabolomic techniques with broader metabolite coverage will be required to investigate the potential for subsite-specific alterations. Some limitations should, however, be acknowledged. First, differences in fasting status between patients and controls represent a potential source of pre-analytical variability. Second, the independent validation cohort lacked matched healthy controls, precluding formal external validation of classifier specificity and overall performance. Consequently, this cohort was used for descriptive validation only. Third, unmeasured factors such as undiagnosed metabolic conditions and recent dietary or drug intake may have influenced metabolite levels. Although exogenous drug-related features were excluded from downstream analyses, the potential impact of unrecorded medication or supplement use on endogenous metabolite levels cannot be completely excluded and therefore represents a common challenge in human metabolomics studies where complete control of medication and supplement use is often not feasible. Future studies with well-characterized cohorts are required to confirm the robustness and clinical utility of the identified markers. Finally, as metabolite annotations were assigned at MSI level-2, structural ambiguity may remain for certain metabolites, particularly for isomeric compounds that share identical mass-to-charge ratios and similar fragmentation patterns. Definitive confirmation would require comparison with authentic chemical standards or the use of additional orthogonal analytical approaches. Nevertheless, our results reinforce the intricate connection between cancer, oxidative stress and glucose dysregulation and suggest a potential association between α-hydroxybutyric acid and metastasis. A deeper understanding of the relationships between glucose dysregulation and cancer could pave the way for improved prevention strategies, prognostic assessments, and novel therapeutic interventions in oral cancer management. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 (973KB, docx) Supplementary Material 2 (237.8KB, xlsx) Supplementary Material 3 (16.5KB, docx) Supplementary Material 4 (21.5KB, docx) Acknowledgements Swedish Metabolomics Centre, Umeå, Sweden (www.swedishmetabolomicscentre.se) is acknowledged for metabolic profiling by GC-MS. Author contributions XG conducted experiments, analyzed data and wrote the manuscript. PJC interpreted data and contributed to manuscript writing. LW analyzed data. NS and MM interpreted data. KN conceived and designed the research and contributed to manuscript writing. All authors read and approved the final manuscript. Funding Open access funding provided by Umea University. This study was supported by Lion’s Cancer Research Foundation, Umeå University; The Swedish Cancer Society (contract number 23 2775 Pj 01 H); Umeå University; Region Västerbotten; Ministry of Health Czech Republic, conceptual development of research organization (MMCI, 00209805). The funding sources had no role other than financial support. Data availability The metabolomic data generated during this study is provided in a supplementary file. Declarations Conflict of interest The authors declare no competing interests. Ethical approval This study was approved by the Regional Ethics Review Board, Umeå, Sweden (Dnr 08-003 M, 2012-131-33 M and 2024-00725-02) and performed in accordance with the Declaration of Helsinki. Written informed consent to participate was obtained from all study subjects. Research involving human participants All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent Informed consent was obtained from all individual participants included in the study. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Alapati, S., Fortuna, G., Ramage, G., & Delaney, C. (2023). Evaluation of Metabolomics as Diagnostic Targets in Oral Squamous Cell Carcinoma: A Systematic Review. Metabolites , 13 (8). 10.3390/metabo13080890 Chen, Y., Wang, B., Zhao, Y., Shao, X., Wang, M., Ma, F., Yang, L., Nie, M., Jin, P., Yao, K., Song, H., Lou, S., Wang, H., Yang, T., Tian, Y., Han, P., & Hu, Z. (2024). Metabolomic machine learning predictor for diagnosis and prognosis of gastric cancer. Nature Communications , 15 (1), 1657. 10.1038/s41467-024-46043-y [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Chicco, D., & Jurman, G. (2020). The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. Bmc Genomics , 21 (1), 6. 10.1186/s12864-019-6413-7 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Danzi, F., Pacchiana, R., Mafficini, A., Scupoli, M. T., Scarpa, A., Donadelli, M., & Fiore, A. (2023). To metabolomics and beyond: a technological portfolio to investigate cancer metabolism. Signal Transduction and Targeted Therapy , 8 (1), 137. 10.1038/s41392-023-01380-0 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Dev, R., Bruera, E., & Dalal, S. (2018). Insulin resistance and body composition in cancer patients. Annals of Oncology , 29 Suppl 2 , ii18–ii26. 10.1093/annonc/mdx815 [ DOI ] [ PubMed ] [ Google Scholar ] Dungan, K. M., Braithwaite, S. S., & Preiser, J. C. (2009). Stress hyperglycaemia. Lancet , 373 (9677), 1798–1807. 10.1016/S0140-6736(09)60553-5 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Enomoto, Y., Kimoto, A., Suzuki, H., Nishiumi, S., Yoshida, M., & Komori, T. (2018). Exploring a Novel Screening Method for Patients with Oral Squamous Cell Carcinoma: A plasma Metabolomics Analysis. Kobe Journal of Medical Sciences , 64 (1), E26–E35. https://www.ncbi.nlm.nih.gov/pubmed/30282895 [ PMC free article ] [ PubMed ] [ Google Scholar ] Ewald, J. D., Zhou, G., Lu, Y., Kolic, J., Ellis, C., Johnson, J. D., Macdonald, P. E., & Xia, J. (2024). Web-based multi-omics integration using the Analyst software suite. Nature Protocols , 19 (5), 1467–1497. 10.1038/s41596-023-00950-4 [ DOI ] [ PubMed ] [ Google Scholar ] Fendt, S. M. (2024). 100 years of the Warburg effect: A cancer metabolism endeavor. Cell , 187 (15), 3824–3828. 10.1016/j.cell.2024.06.026 [ DOI ] [ PubMed ] [ Google Scholar ] Finley, L. W. S. (2023). What is cancer metabolism? Cell , 186 (8), 1670–1688. 10.1016/j.cell.2023.01.038 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Gall, W. E., Beebe, K., Lawton, K. A., Adam, K. P., Mitchell, M. W., Nakhle, P. J., Ryals, J. A., Milburn, M. V., Nannipieri, M., Camastra, S., Natali, A., Ferrannini, E., & Grp, R. S. (2010). α-Hydroxybutyrate Is an Early Biomarker of Insulin Resistance and Glucose Intolerance in a Nondiabetic Population. PloS One , 5 (5), e10883. 10.1371/journal.pone.0010883 Giri, B., Dey, S., Das, T., Sarkar, M., Banerjee, J., & Dash, S. K. (2018). Chronic hyperglycemia mediated physiological alteration and metabolic distortion leads to organ dysfunction, infection, cancer progression and other pathophysiological consequences: An update on glucose toxicity. Biomedicine and Pharmacotherapy , 107 , 306–328. 10.1016/j.biopha.2018.07.157 [ DOI ] [ PubMed ] [ Google Scholar ] Greener, J. G., Kandathil, S. M., Moffat, L., & Jones, D. T. (2022). A guide to machine learning for biologists. Nature Reviews: Molecular Cell Biology , 23 (1), 40–55. 10.1038/s41580-021-00407-0 [ DOI ] [ PubMed ] [ Google Scholar ] Gullberg, J., Jonsson, P., Nordstrom, A., Sjostrom, M., & Moritz, T. (2004). Design of experiments: an efficient strategy to identify factors influencing extraction and derivatization of Arabidopsis thaliana samples in metabolomic studies with gas chromatography/mass spectrometry. Analytical Biochemistry , 331 (2), 283–295. 10.1016/j.ab.2004.04.037 [ DOI ] [ PubMed ] [ Google Scholar ] Gupta, I., Badrzadeh, F., Tsentalovich, Y., & Gaykalova, D. A. (2024). Connecting the dots: investigating the link between environmental, genetic, and epigenetic influences in metabolomic alterations in oral squamous cell carcinoma. Journal of Experimental and Clinical Cancer Research , 43 (1), 239. 10.1186/s13046-024-03141-5 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Hardie, D. G. (2022). 100 years of the Warburg effect: a historical perspective. Endocrine-Related Cancer , 29 (12), T1–T13. 10.1530/Erc-22-0173 [ DOI ] [ PubMed ] [ Google Scholar ] Irajizad, E., Kenney, A., Tang, T., Vykoukal, J., Wu, R., Murage, E., Dennison, J. B., Sans, M., Long, J. P., Loftus, M., Chabot, J. A., Kluger, M. D., Kastrinos, F., Brais, L., Babic, A., Jajoo, K., Lee, L. S., Clancy, T. E., Ng, K., & Fahrmann, J. F. (2023). A blood-based metabolomic signature predictive of risk for pancreatic cancer. Cell Reports Medicine , 4 (9), 101194. 10.1016/j.xcrm.2023.101194 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Kanter, N. G., Cohen-Woods, S., Balfour, D. A., Burt, M. G., Waterman, A. L., & Koczwara, B. (2024). Hypothalamic-Pituitary-Adrenal Axis Dysfunction in People With Cancer: A Systematic Review. Cancer Medicine , 13 (22), e70366. 10.1002/cam4.70366 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Kong, X., Yang, X., Zhou, J., Chen, S., Li, X., Jian, F., Deng, P., & Li, W. (2015). Analysis of plasma metabolic biomarkers in the development of 4-nitroquinoline-1-oxide-induced oral carcinogenesis in rats. Oncology Letters , 9 (1), 283–289. 10.3892/ol.2014.2619 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Lapenna, D. (2023). Glutathione and glutathione-dependent enzymes: From biochemistry to gerontology and successful aging. Ageing Research Reviews , 92 , 102066. 10.1016/j.arr.2023.102066 [ DOI ] [ PubMed ] [ Google Scholar ] Lord, R. S., & Bralley, J. A. (2008). Clinical applications of urinary organic acids. Part I: Detoxification markers. Alternative Medicine Review , 13 (3), 205–215. https://www.ncbi.nlm.nih.gov/pubmed/18950247 [ PubMed ] [ Google Scholar ] Luo, G., Wang, S., Lu, W., Ju, W., Li, J., Tan, X., Zhao, H., Han, W., & Yang, X. (2024). Application of metabolomics in oral squamous cell carcinoma. Oral Diseases , 30 (6), 3719–3731. 10.1111/odi.14895 [ DOI ] [ PubMed ] [ Google Scholar ] Lv, X., Ding, P., Li, L., Li, L., Zhou, D., Wang, X., Chen, J., Zhang, W., Wang, Q., Liao, T., Wen, W., Zhou, D., Ji, Q. H., He, X., Lei, Q. Y., & Hu, W. (2023). Increased alpha-HB links colorectal cancer and diabetes by potentiating NF-kappaB signaling. Molecular Metabolism , 75 , 101766. 10.1016/j.molmet.2023.101766 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Marmol, J. M., Carlsson, M., Raun, S. H., Grand, M. K., Sorensen, J., Lang Lehrskov, L., Richter, E. A., Norgaard, O., & Sylow, L. (2023). Insulin resistance in patients with cancer: a systematic review and meta-analysis. Acta Oncologica , 62 (4), 364–371. 10.1080/0284186X.2023.2197124 [ DOI ] [ PubMed ] [ Google Scholar ] Masi, T., & Patel, B. M. (2021). Altered glucose metabolism and insulin resistance in cancer-induced cachexia: a sweet poison. Pharmacological Reports , 73 (1), 17–30. 10.1007/s43440-020-00179-y [ DOI ] [ PubMed ] [ Google Scholar ] Ogawa, T., Washio, J., Takahashi, T., Echigo, S., & Takahashi, N. (2014). Glucose and glutamine metabolism in oral squamous cell carcinoma: insight from a quantitative metabolomic approach. Oral Surg Oral Med Oral Pathol Oral Radiol , 118 (2), 218–225. 10.1016/j.oooo.2014.04.003 [ DOI ] [ PubMed ] [ Google Scholar ] Pang, Z., Lu, Y., Zhou, G., Hui, F., Xu, L., Viau, C., Spigelman, A. F., MacDonald, P. E., Wishart, D. S., Li, S., & Xia, J. (2024). MetaboAnalyst 6.0: towards a unified platform for metabolomics data processing, analysis and interpretation. Nucleic Acids Research , 52 (W1), W398–W406. 10.1093/nar/gkae253 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Patra, K. C., & Hay, N. (2014). The pentose phosphate pathway and cancer. Trends in Biochemical Sciences , 39 (8), 347–354. 10.1016/j.tibs.2014.06.005 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Polachini, G. M., de Castro, T. B., Smarra, L. F. S., Henrique, T., de Paula, C. H. D., Severino, P., Lopez, R. V. M., Carvalho, A. L., de Mattos Zeri, A. C., Silva, I., & Tajara, E. H. (2023). Plasma metabolomics of oral squamous cell carcinomas based on NMR and MS approaches provides biomarker identification and survival prediction. Scientific Reports , 13 (1), 8588. 10.1038/s41598-023-34808-2 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Schauer, N., Steinhauser, D., Strelkov, S., Schomburg, D., Allison, G., Moritz, T., Lundgren, K., Roessner-Tunali, U., Forbes, M. G., Willmitzer, L., Fernie, A. R., & Kopka, J. (2005). GC-MS libraries for the rapid identification of metabolites in complex biological samples. FEBS Letters , 579 (6), 1332–1337. 10.1016/j.febslet.2005.01.029 [ DOI ] [ PubMed ] [ Google Scholar ] Schmidt, D. R., Patel, R., Kirsch, D. G., Lewis, C. A., Vander Heiden, M. G., & Locasale, J. W. (2021). Metabolomics in cancer research and emerging applications in clinical oncology. CA: A Cancer Journal for Clinicians , 71 (4), 333–358. 10.3322/caac.21670 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Song, X., Yang, X., Narayanan, R., Shankar, V., Ethiraj, S., Wang, X., Duan, N., Ni, Y. H., Hu, Q., & Zare, R. N. (2020). Oral squamous cell carcinoma diagnosed from saliva metabolic profiling. Proceedings of the National Academy of Sciences of the United States of America , 117 (28), 16167–16173. 10.1073/pnas.2001395117 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Sousa, A. P., Cunha, D. M., Franco, C., Teixeira, C., Gojon, F., Baylina, P., & Fernandes, R. (2021). Which Role Plays 2-Hydroxybutyric Acid on Insulin Resistance? Metabolites , 11 (12). 10.3390/metabo11120835 Suhre, K., Stephan, N., Zaghlool, S., Triggle, C. R., Robinson, R. J., Evans, A. M., & Halama, A. (2022). Matching Drug Metabolites from Non-Targeted Metabolomics to Self-Reported Medication in the Qatar Biobank Study. Metabolites , 12 (3). 10.3390/metabo12030249 Swanton, C., Bernard, E., Abbosh, C., Andre, F., Auwerx, J., Balmain, A., Bar-Sagi, D., Bernards, R., Bullman, S., DeGregori, J., Elliott, C., Erez, A., Evan, G., Febbraio, M. A., Hidalgo, A., Jamal-Hanjani, M., Joyce, J. A., Kaiser, M., Lamia, K., & Hanahan, D. (2024). Embracing cancer complexity: Hallmarks of systemic disease. Cell , 187 (7), 1589–1616. 10.1016/j.cell.2024.02.009 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Tan, Y., Wang, Z., Xu, M., Li, B., Huang, Z., Qin, S., Nice, E. C., Tang, J., & Huang, C. (2023). Oral squamous cell carcinomas: state of the field and emerging directions. International Journal Of Oral Science , 15 (1), 44. 10.1038/s41368-023-00249-w [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Tasdogan, A., Ubellacker, J. M., & Morrison, S. J. (2021). Redox Regulation in Cancer Cells during Metastasis. Cancer Discovery , 11 (11), 2682–2692. 10.1158/2159-8290.CD-21-0558 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Tiziani, S., Lopes, V., & Gunther, U. L. (2009). Early stage diagnosis of oral cancer using 1H NMR-based metabolomics. Neoplasia (New York, N.Y.) , 11 (3), 269–276. 10.1593/neo.81396. 264p following 269. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Tsai, C. K., Lin, C. Y., Kang, C. J., Liao, C. T., Wang, W. L., Chiang, M. H., Yen, T. C., & Lin, G. (2020). Nuclear Magnetic Resonance Metabolomics Biomarkers for Identifying High Risk Patients with Extranodal Extension in Oral Squamous Cell Carcinoma. Journal of Clinical Medicine , 9 (4). 10.3390/jcm9040951 Vitorio, J. G., Duarte-Andrade, F. F., Santos Fontes Pereira, D., Fonseca, T., Amorim, F. P., Martins-Chaves, L. S. D., Gomes, R. R., Canuto, C. C., G. A. B., & Gomez, R. S. (2020). Metabolic landscape of oral squamous cell carcinoma. Metabolomics , 16 (10), 105. 10.1007/s11306-020-01727-6 [ DOI ] [ PubMed ] [ Google Scholar ] Wang, W., Rong, Z., Wang, G., Hou, Y., Yang, F., & Qiu, M. (2023). Cancer metabolites: promising biomarkers for cancer liquid biopsy. Biomarker Research , 11 (1), 66. 10.1186/s40364-023-00507-3 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Wishart, D. S. (2016). Emerging applications of metabolomics in drug discovery and precision medicine. Nature Reviews: Drug Discovery , 15 (7), 473–484. 10.1038/nrd.2016.32 [ DOI ] [ PubMed ] [ Google Scholar ] Yang, X. H., Song, X. W., Yang, X. D., Han, W., Fu, Y., Wang, S., Zhang, X. X., Sun, G. W., Lu, Y., Wang, Z. Y., Ni, Y. H., Zare, R. N., & Hu, Q. G. (2022). Big cohort metabolomic profiling of serum for oral squamous cell carcinoma screening and diagnosis. Natural Sciences , 2 (1). 10.1002/ntls.20210071 Zeng, J., Yin, P., Tan, Y., Dong, L., Hu, C., Huang, Q., Lu, X., Wang, H., & Xu, G. (2014). Metabolomics study of hepatocellular carcinoma: discovery and validation of serum potential biomarkers by using capillary electrophoresis-mass spectrometry. Journal of Proteome Research , 13 (7), 3420–3431. 10.1021/pr500390y [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary Material 1 (973KB, docx) Supplementary Material 2 (237.8KB, xlsx) Supplementary Material 3 (16.5KB, docx) Supplementary Material 4 (21.5KB, docx) Data Availability Statement The metabolomic data generated during this study is provided in a supplementary file. Articles from Metabolomics are provided here courtesy of Springer ACTIONS View on publisher site PDF (1.2 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top