Integration RNA bulk and single cell RNA sequencing to explore the change of BCAA metabolism-related immune microenvironment and construct prognostic signature in HNSCC

Integration RNA bulk and single cell RNA sequencing to explore the change of BCAA metabolism-related immune microenvironment and construct prognostic signature in HNSCC | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Integration RNA bulk and single cell RNA sequencing to explore the change of BCAA metabolism-related immune microenvironment and construct prognostic signature in HNSCC Dandan Lu, Yongjun Liang, Tao Mo, Abdeyrim Arikin This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6030799/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 Several studies have demonstrated that impaired metabolism of branched chain amino acids (BCAAs) is related to cancer progression. However, the specific mechanisms underlying BCAA metabolism in head and neck squamous cell carcinoma (HNSCC) remain to be explored. The aim of this study was to identify prognostic genes associated with BCAA metabolism in HNSCC and to elucidate their functional mechanisms. Methods The HNSCC related datasets (TCGA-HNSCC, GSE65858 and GSE140042) were enrolled in this study. Candidate genes were acquired by overlapping differentially expressed genes form differential expression analysis and key module genes connected with BCAA-metabolism related genes (BCAA-MRGs) scores from weighted gene co-expression network analysis. Subsequently, prognostic genes were obtained to construct the risk model through univariate Cox regression analysis, proportional hazards hypothesis test, and least absolute shrinkage and selection operator regression analysis selected in sequence. Afterwards, independent prognostic analysis, enrichment analysis and immune microenvironment analysis were performed. Furthermore, the expression changes of prognostic genes at the cellular level were assessed through single-cell RNA sequencing (scRNA-seq) data analysis and pseudo-time analysis. Additionally, RT-qPCR was used to confirm the expression levels of prognostic genes in HNSCC tissues. Results SMS, PRDX6, GSTO1, and ADA were determined as prognostic genes to create the risk model. The HNSCC samples were divided into high-risk group (HRG) and low-risk group (LRG), with LRG demonstrating significantly higher survival rates compared to the HRG. Furthermore, the nomogram model constructed using risk score and age had an excellent predictive ability for HNSCC patients. Enrichment analysis revealed that ‘pentose phosphate pathway’ and ‘fructose and mannose metabolism’ were significantly associated with HNSCC progression. At the same time, we also found that the level of infiltration of 20 immune cells (plasmacytoid dendritic cells, mast cells, and T follicle helper cells) and the expression of 10 immune checkpoints (CD276, CD27, and CD40) differed between the HRG and the LRG. Additionally, epithelial cells were selected as key cells due to higher expression of prognostic genes. Importantly, the trend of prognostic gene expression varied with different stages of cell differentiation. Through RT-qPCR experiment, SMS, GSTO1, and ADA all expressed highly in the tumor group, but PRDX6 had not remarkably difference between tumor and normal groups. Conclusion In summary, we pinpointed four genes-SMS, PRDX6, GSTO1, and ADA-linked to the prognosis of HNSCC within the context of BCAA metabolism. Subsequently, we developed a risk model. This model offers a novel reference for prognostic assessment and treatment strategies tailored to HNSCC patients. Biological sciences/Cancer Biological sciences/Computational biology and bioinformatics Biological sciences/Drug discovery Biological sciences/Immunology Health sciences/Biomarkers Head and neck squamous cell carcinoma branched chain amino acids metabolism Single-cell RNA sequencing Epithelial cells Risk model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Introduction Head and neck tumors have the seventh highest incidence rate of all cancers in the 2022 National Cancer Statistics 1 . HNSCC is the most common pathologic type of head and neck tumors. Despite impressive advances in surgery, radiation, chemotherapy, targeted therapies, and especially immunotherapy in recent years, the prognosis for HNSCC is dishearteningly poor, as the 5-year survival rate hovers at merely around 50% 2 . In addition, due to the heterogeneity of HNSCC, histopathologic and clinical staging are not sufficient to accurately predict patient prognosis 3 . Therefore, it is of utmost urgency to identify new targets for prognostic prediction and treatment of HNSCC. Abnormal energy metabolism is a fundamental feature of tumors, in which branched-chain amino acid (BCAA, including leucine, valine, isoleucine) metabolism plays a significance role in tumor progression 4 – 6 . Cancer cells reshape their metabolism to evade immune surveillance in the tumor microenvironment (TME), forming nutrient-depleted TME and dysfunctional T cells 7 , 8 . Depletion of BCAAs in TME may affect the ability of tumor infiltrating lymphocytes (TILs) to eliminate cancer cells. In the past, the absence of anticipated outcomes from tumor immunotherapy has been intimately linked to the suppressive impact of metabolic reprogramming on immune cells in TME 9 . Lymphocyte regulation is influenced by amino acid metabolism. A strong correlation between BCAA intake and total lymphocyte count (TLC) was found in patients with HNSCC in the underweight and elderly groups of the study 10 . Several in vitro studies have shown that BCAA deficiency leads to impaired lymphocyte proliferation 11 , 12 . BCAA metabolism not only enhances energy provision but also actively promotes immune escape of tumor cells 13 . Leucine is an indispensable substance for the activation of T cells, as a deficiency in leucine inhibits the clonal expansion of Th1, Th17, and CD8 + T cell 14 . Leucine has been discovered to exert an influence on the mammalian target of rapamycin (mTOR) signaling pathway. This pathway is not merely pivotal in orchestrating T-cell differentiation and function; it also wields control over protein translation, as well as cell growth and proliferation 15 , 16 . BCAA metabolism is associated with the development of many cancers, including breast cancers 5 , 17 , glioblastoma 18 , colorectal cancer 19 , and renal cancer 6 . It has also been shown that BCAA-MRGs play an indispensable part in prognosticating the outcome of a wide variety of tumors, including multiple myeloma 20 , renal clear cell carcinoma 21 , and pancreatic cancer 22 . In summary, BCAA-MRGs show prognostic value in a variety of cancers, demonstrating their great potential for personalized therapy. To date, research on BCAA metabolism in HNSCC has been limited. This study explores the potential of BCAA-MRGs as prognostic genes for HNSCC and their possible molecular mechanisms through bioinformatics analysis. Furthermore, single-cell RNA sequencing (scRNA-seq) is used to assess the expression of prognostic genes at the cellular level, and in vitro experiments are conducted to validate the functions of BCAA-MRGs. Materials and methods Data extraction The Cancer Genome Atlas (TCGA) database ( https://gdc.cancer.gov/ ) was applied to obtain the gene expression matrix and clinical data of TCGA-HNSCC dataset, which comprised 502 HNSCC tissue samples and 44 paracancerous tissue samples. Of these, 501 HNSCC samples had complete survival information. This dataset was used as a training set. On the other hand, GSE65858 and GSE140042 datasets were accessed through Gene Expression Omnibus (GEO) database ( http://www.ncbi.nlm.nih.gov/geo/ ). The GSE65858 (platform: GPL10558) dataset was utilized as a validation set, encompassing gene expression matrices and clinical data derived from tumor tissues of 270 HNSCC patients. The scRNA-seq dataset GSE140042 (platform: GPL16791) included tumor tissue samples obtained from 10 HNSCC patients. Additionally, a total of 27 BCAA-MRGs were selected from Molecular Signatures Database (MSigDB) ( https://www.gsea-msigdb.org/ ) with ‘branched chain amino acids metabolism’ as a keyword. Differential expression analysis Differential expression analysis was carried out using the DEseq2 package 23 (v 3.4.1) to pinpoint differentially expressed genes (DEGs) between tumor and normal tissues within the TCGA-HNSCC dataset. The P 0.5 were utilized as screening criteria. With the application of ggplot2 24 (v 3.4.1) and ComplexHeatmap packages 25 (v 2.14.0), volcano plot and heat map were generated to reveal the expression of DEGs, respectively. Specifically, the heat map exhibited the top10 up-regulated and down-regulated DEGs. Weighted gene co-expression network analysis (WGCNA) The BCAA-MRGs scores for 501 HNSCC samples with complete survival information in the TCGA-HNSCC dataset were computed employing the single-sample gene set enrichment analysis (ssGSEA) algorithm in the GSVA package 26 (v 1.46.0). Meanwhile, 501 HNSCC samples with complete survival information were divided into high and low score groups according to the best cutoff value for BCAA-MRGs scores, and then Kaplan-Meier (K-M) survival analyses were performed using the survminer software package 27 (v 0.4.9) (log-rank test, P < 0.05) in order to investigate whether there was a difference in survival between the two score groups. WGCNA was conducted using WGCNA package 28 (v 1.70.3) in the 501 HNSCC samples with complete survival information of TCGA-HNSCC dataset, aiming to gain the gene modules associated with BCAA-MRGs scores. To begin with, the HNSCC sample was subjected to a clustering analysis with the objective of assessing the presence of outliers in the sample that needed to be eliminated. Subsequently, the optimal soft threshold power (β) was determined by a scale-free topological fit index greater than 0.85 and a mean connectivity tending towards 0. Following this, the systematic clustering tree was constructed based on the adjacency connection and gene similarity, followed by the construction of a co-expression network using the hybrid dynamic tree cutting algorithm, with a minimum requirement of 100 genes in each gene module. Afterwards, using BCAA-MRGs scores as phenotypic traits, Pearson correlation analysis was performed to calculate the correlation coefficient between modules and BCAA-MRGs scores. The module with the highest correlation with BCAA-MRGs scores was defined as the key module. Furthermore, the genes with |module membership (MM)| > 0.2 and |gene significance (GS)| > 0.2 in the key module were identified as key module genes highly related to BCAA-MRGs scores. Recognition and functional analysis of candidate genes Candidate genes were acquired by intersecting DEGs and key module genes using VennDiagram package 29 (v 1.7.3). Based on the GO and KEGG databases, the clusterProfiler package 30 (v 4.7.1.3) was utilized for the enrichment analysis of candidate genes with the aim of exploring the biological functions and signaling pathways of candidate genes in the pathogenesis of HNSCC ( P .adjust < 0.05). Furthermore, to gain insight into the interactions between candidate genes, these genes were introduced into the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) ( http://string-db.org ) website to construct a protein-protein interaction (PPI) network (interaction score > 0.4), which was then visualized using Cytoscape software 31 (v 3.7.2) for visualization. Screening for prognostic genes In the 501 HNSCC samples with complete survival information within TCGA-HNSCC dataset, univariate Cox regression analysis was implemented via the survival package 32 (v 3.3-1), resulting in genes significantly associated with the survival of HNSCC [hazard ratio (HR) ≠ 1 & P < 0.05]. Subsequently, the forest map was generated for presenting this result using forestplot package 33 (v 2.0.1). Following this, the proportional hazards (PH) hypothesis test was implemented to target these genes that were significantly associated with HNSCC survival ( P > 0.05). Subsequently, the genes that passed the PH hypothesis test were incorporated into the least absolute shrinkage and selection operator (LASSO) regression analysis. This analysis was executed using the glmnet package 34 (v 4.1.4) to conduct further screening, with the aim of identifying prognostic genes associated with HNSCC. Besides, the expression levels of prognostic genes between tumor and normal groups in TCGA-HNSCC dataset were assessed by the application of the Wilcoxon test ( P < 0.05). Construction and verification of risk model The risk model was constructed based on the identified prognostic genes in the 501 HNSCC samples with complete survival information within TCGA-HNSCC dataset. The risk score was calculated as follows: riskscore = In this formula, βi denoted LASSO regression coefficient, while Expri represented the expression of prognostic gene i. The HNSCC samples were divided into high risk group (HRG) and low risk group (LRG) based on the best cut-off value of risk score. Subsequently, risk curves and survival status maps were plotted, aiming to demonstrate the distribution of patients with HNSCC. Immediately thereafter, survival differences between HRG and LRG were further compared by K-M survival analysis with survminer package (log-rank test, P < 0.05). Furthermore, receiver operating characteristic (ROC) curves were constructed using the survival ROC package 35 (v 1.0.3). Subsequently, area under the curve (AUC) values were calculated. This was done to elucidate the predictive effect of the risk model on the prognosis of HNSCC patients. Importantly, the predictive performance of the risk model was verified in the GSE65858 dataset with the application of risk curves and survival status maps, K-M survival analysis, and ROC curves. Independent prognostic analysis and nomogram evaluation Univariate and multivariate Cox regression analyses ( P < 0.05) were implemented for risk score and clinical features (stage, age, gender, T and N stage) to select the independent prognostic factors in the TCGA-HNSCC dataset. After that, a nomogram was constructed employing rms package 36 (v 6.5.0) based on the identified independent prognostic factors with the aim of predicting 1/3/5-year survival in patients with HNSCC. Furthermore, the calibration curves and ROC curves were constructed to assess the predictive capabilities of the nomogram model. Clinical characteristics analysis Studies have divided all tumors in the TCGA dataset into 6 immune subtypes, namely wound healing (C1), IFN-γ dominant (C2), inflammatory (C3), lymphocyte depleted (C4), immunologically quiet (C5), and TGF-β dominant (C6) 37 , 38 . To comprehend the immune subtypes and stage between HRG and LRG, the RColorBrewer package 39 (v 1.1.3) was employed for analyzing the significance of immune subtypes and stage in the TCGA-HNSCC dataset. Additionally, ROC curve analysis was conducted to assess the prognostic predictive ability of risk scores and clinical features for HNSCC patients. Functional enrichment analysis In order to elucidate the biological functional disparities between HRG and LRG of HNSCC patients in TCGA-HNSCC dataset, differential expression analysis was conducted using the DEseq2 package, with c2.kegg.symbols.gmt from MSigDB serving as a reference gene set. Subsequently, genes were ranked based on their calculated log2FC values. Following this step, gene set enrichment analysis (GSEA) was performed utilizing the clusterProfiler package. The screening criteria were defined as |normalized enrichment scores (NES)| > 1, false discovery rate (FDR) < 0.05 and P < 0.05. Furthermore, gene set variation analysis (GSVA) was implemented via GSVA package to mine the pathway differences between HRG and LRG, with c2.kegg.symbols.gmt as the reference gene set. Additionally, univariate Cox regression analysis was performed using the survival package to explore the relationship between top 5 pathways obtained from GSVA and survival in HNSCC patients (HR ≠ 1 & P < 0.05). Immune microenvironment analysis To estimate the TME in patients, we initially computed the stromal score, immune score, and ESTIMATE score for each sample from the TCGA-HNSCC dataset using the ESTIMATE algorithm. Subsequently, we employed the Wilcoxon test to compare any discrepancies in these three scores between HRG and LRG ( P < 0.05). The infiltration scores of the 28 immune cells in each sample were calculated using the ssGSEA algorithm from the GSVA package, and a comparison was made between HRG and LRG by conducting a Wilcoxon test to assess differences in infiltration scores among these immune cells ( P < 0.05). Afterwards, Spearman correlation analysis was concluded to reveal the relationship between top five differential immune cells identified according to the significance of the infiltration level and risk score, as well as between differential immune cells and prognostic genes using psych package 40 (v 2.2.9) [|correlation coefficient (cor)|>0.3 & P < 0.05]. Furthermore, HNSCC patients were categorized into high- and low-expression groups based on the optimal cut-off values for differential immune cell expression levels, followed by implementing K-M survival analysis to evaluate survival differences between the high- and low-expression groups (log-rank test, P < 0.05). On the other hand, the expression of 24 immune checkpoints was also evaluated between HRG and LRG through Wilcoxon test 41 ( P < 0.05). Additionally, the TIDE score, dysfunction score, and exclusion score for HNSCC samples between HRG and LRG were compared by Wilcoxon test ( P < 0.05). Somatic mutation analysis In the TCGA-HNSCC dataset, somatic mutations of patients in HRG and LRG were analyzed using the maftools package 42 (v 2.14.0). In order to evaluate the association between tumor mutational burden (TMB) and patient survival, HNSCC patients within the TCGA-HNSCC dataset were classified into high-TMB and low-TMB groups. This categorization was based on the median value of TMB. Subsequently, K-M survival analysis was employed to explore the survival differences in HNSCC patients among the HRG + high TMB group, HRG + low TMB group, LRG + high TMB group, and LRG + low TMB group. Drug sensitivity analysis To identify potential therapeutic drugs for patients in HRG, the oncoPredict package 43 (v 0.2) was employed to predict potential therapeutic drugs in the TCGA-HNSCC dataset. Subsequently, Spearman correlation analysis was performed to assess the association between risk score and the AUC values of these drugs (|cor|>0.3 & P < 0.05). scRNA-seq data analysis The expression levels of prognostic genes at the cellular level were assessed in GSE140042 dataset. First, the scRNA-seq data were converted to Seurat objects using the Seurat package 44 (v 5.0.1). The scRNA-seq data were then filtered using the ‘CreateSeuratObject’ function of the Seurat package, retaining cells with the number of detected genes above 200, followed by calculating the mitochondrial genes using the ‘PercentageFeatureSet’ function, and retaining cells with a mitochondrial gene percentage of less than 5%. After excluding cells and genes that did not meet the criteria, the data were normalized using the ‘NormalizeData’ function from the Seurat package with the following parameters: logNormalize, scale.factor = 10000. Then, the ‘vst’ method from ‘FindVariableFeatures’ function was employed to extract the genes with relatively high coefficients of variation between cells, resulting in the need to find the top 2,000 highly variable genes in this analysis. Following this, principal component analysis (PCA) was examined for the presence of outliers or abnormal cells, followed by the ‘JackStrawPlot’ function to identify statistically significant principal components (PC) ( P < 0.05). The ‘NormalizeData’ function was utilized to normalize the data for further clustering of cell clusters. Afterwards, unsupervised cluster analysis was completed using ‘FindNeighbors’ and ‘FindClusters’ functions to yield cell clusters (resolution = 1). Based on the cell clusters obtained from the above clustering analysis, the marker genes of each cell type were identified by the ‘FindAllMarkers’ function with the parameters set to only.pos = TRUE and logfc.threshold = 1. In conjunction with the auxiliary annotation provided by the ‘SingleR’ function, the cell clusters were annotated using the related marker genes obtained in the published literature 45 to identify different cell types. Pseudo-time analysis for key cells The expression levels of prognostic genes were further compared in different cell types, and then key cells with higher prognostic gene expression were identified. Subsequently, the key cells were subjected to dimensionality reduction and clustering analysis to identify different subpopulations of key cells using the same analysis method as the scRNA-seq data analysis described above. The FindAllMarkers setting parameters min.pct = 0.5, logfc.threshold = 0.25, test.use = auc were employed to find the marker genes for each subpopulation. Immediately thereafter, the marker genes for each cell subpopulation were compared with the marker genes for each cell type in the CellMarker database to determine the subpopulation types of key cells. To explore the state changes of key cells and the expression changes of significantly changed genes, pseudo-time analysis of key subpopulation cells was first performed using Monocle2 package 46 (v 2.22.0). Then, the pseudo-time trajectory was inferred using the ‘reduceDimension’ function with its default parameters, and the reconstructed trajectory was visualized by utilizing the ‘plot cell trajectory’ function. Afterwards, 2,000 highly variable genes were extracted, and the genes were clustered according to the pseudo-time expression pattern to generate a representative heat map. Furthermore, the expression of prognostic genes in different developmental states was further analyzed. Reverse transcription quantitative polymerase chain reaction (RT-qPCR) In an attempt to validate the expression of prognostic genes, 10 HNSCC cancer samples and 10 paracancerous tissue samples were collected from People’s Hospital of Xinjiang Uygur Autonomous Region. First, RNA was extracted using 1 ml TRIzol reagent with 50 mg tissue homogenate from each sample, and its concentration was quantified using a NanoPhotometer N50. Then, cDNA was synthesized by reverse transcription according to the SureScript-First-strand-cDNA-synthesis-kit instructions provided by Xavier's. The cDNA was then amplified for 40 cycles using the CFX96 real-time fluorescence quantitative PCR instrument. For RT-qPCR analysis, there were 3 µL cDNA, 5 µL 2x Universal Blue SYBR Green qPCR Master Mix (Servicebio), and 1 µL specific primers per reaction mix. Primer sequences were shown in Supplementary Table 1 . Finally, expression of prognostic genes was assessed using the 2-ΔΔCt method. Statistical analysis Statistical analyses in this study were conducted utilizing R software (v 4.2.3). Differences between the two groups were compared through the implementation of Wilcoxon test. The P value less than 0.05 was considered statistically significant. Results Recognition of 8,533 DEGs and 1,369 key module genes A total of 8,533 DEGs (tumor vs. normal) were selected through differential expression analysis. Among these, 4,360 genes were found to be up-regulated, while 4,173 genes were down-regulated specifically in the tumor group (Fig. 1 A-B). The Wilcoxon test results revealed differences in BCAA-MRG scores between the tumor and normal groups, with significantly higher scores observed in the normal group. Furthermore, K-M survival analysis between the high and low score groups divided according to the optimal cut-off value of this score demonstrated a significantly worse prognosis for the low score group compared to the high score group ( P < 0.0084) (Fig. 1 C). These findings revealed that BCAA-MRG scores were able to serve as phenotypic traits to implement WGCNA. The clustering analysis results indicated that there were no outliers among the 501 HNSCC samples, so all of these samples were included in the subsequent analyses (Fig. 1 D). Then, the optimal soft threshold power (β) was determined to be four based on a scale-free topological fit index greater than 0.85 and a mean connectivity tending towards 0 (Fig. 1 E). Following this, 11 gene modules were obtained according to the criteria for the construction of a co-expression network (Fig. 1 F). Among these gene modules, MEyellow (cor=-0.7288, P = 1.98×10 − 20 ) presented the most significant correlation with BCAA-MRG scores and was selected as the key module (Fig. 1 G). Importantly, 1,369 key module genes were gained based on the screening criteria of |MM| > 0.2 and |GS| > 0.2 (Fig. 1 H). Multiple functions of 257 candidate genes with interactions were characterized A total of 257 candidate genes were identified by crossing 8,533 DEGs and 1,369 key module genes (Fig. 2 A). The signaling biological functions and pathways in which these candidate genes were significantly enriched were further explored by enrichment analyses. The results revealed that these candidate genes were significantly enriched to 146 GO entries, which contained 26 molecular functions, 70 cellular components, and 50 biological processes, as well as 31 KEGG pathways. According to the P .adjust value ranking, the enriched top10 GO entries (Fig. 2 B) and top10 KEGG pathways (Fig. 2 C) were selected for presentation, such as aerobic respiration, ATP synthesis coupled electron transport, Chemical carcinogenesis-reactive oxygen species, oxidative phosphorylation, respiratory electron transport chain, thermogenesis, etc. These results implicated that the candidate genes might be influencing the development of HNSCC by affecting the energy metabolism and conversion of cells. Additionally, the results of the PPI network demonstrated that UQCRFS1, UQCRC2, UQCRC1, COX7C, NDUFA4, and SNRPE interacted with more genes ( Supplementary Fig. 1 ). SMS, PRDX6, GSTO1, and ADA were confirmed as prognostic genes to construct the risk model Six genes significantly associated with HNSCC survival were identified, namely PFDN4, SMS, PRDX6, GSTO1, COA6, and ADA, employing univariate Cox regression analysis, and all of these genes were risk factors for HNSCC (HR > 1) (Fig. 3 A). Meanwhile, these genes all passed the PH hypothesis test ( P > 0.05) ( Supplementary Fig. 2A ). In the LASSO regression analysis, SMS, PRDX6, GSTO1, and ADA were identified and recorded as prognostic genes (lambda.min = 0.01019419) (Fig. 3 B). Furthermore, the expression assessment of these prognostic genes indicated that PRDX6 was downregulated in the tumor group compared to the normal group, whereas SMS, GSTO1, and ADA were upregulated (Fig. 3 C). Similarly, in the RT-qPCR experiment, SMS, GSTO1, and ADA all expressed highly in the tumor group ( P 0.05) (Fig. 3 G). Based on the expression levels of the four prognostic genes and the corresponding coefficients in LASSO regression analysis, a risk model was constructed utilizing the following formula: risk score = 0.2138×SMS expression level + 0.2144 × PRDX6 expression level + 0.0512 × GSTO1 expression level + 0.1510 × ADA expression level. The HNSCC samples with complete survival information were classified into HRG and LRG based on the best cut-off value of risk score. As demonstrated in Supplementary Fig. 2B , it could be found that the mortality rate of patients increased with the increase of risk score. K-M survival analysis confirmed significant differences in survival between HRG and LRG, with HRG having significantly worse survival compared to LRG ( P = 0.0001) (Fig. 3 H). ROC curve indicated an appropriate accuracy of risk model in predicting survival, with AUC values of 0.615, 0.631, 0.614 at 1-, 3-, and 5-year, respectively (Fig. 3 I). Furthermore, the robust prediction ability of risk model was validated in GSE65858 dataset, yielding consistent results. Specifically, the risk curves and survival status maps revealed that as the risk score increased, the mortality rate of patients also increased. K-M survival analysis confirmed that LRG patients had notably lower survival than HRG patients. Also, the AUC values in the ROC curves were all greater than 0.6 (Fig. 3 J-K, Supplementary Fig. 2C ). Exploration of predictive performance for nomogram and clinical features Two independent prognostic factors, risk score and age, were generated through the application of univariate and multivariate Cox logistic regression analyses (Fig. 4A-B). The nomogram model was constructed using risk score and age to predict 1/3/5-year survival in patients with HNSCC (Fig. 4C). The calibration curves indicated that the slopes of the nomogram model at 1/3/5 years were close to the theoretical line, indicating a high level of predictive accuracy for the nomogram model (Fig. 4D). Meanwhile, the AUC values of the nomogram model were consistently above 0.6 at 1 (0.894), 3 (0.678), and 5 (0.678) years, revealing the excellent predictive ability of the nomogram model for the survival of HNSCC patients (Fig. 4E). On the other hand, the differences of immune subtypes and stage between HRG and LRG were compared. The results revealed that the immune subtypes were not significantly different between HRG and LRG (P = 0.088) ( Supplementary Fig. 3A ), whereas stage was significantly different (P = 0.013) between HRG and LRG ( Supplementary Fig. 3B ). Furthermore, the ROC curve analysis demonstrated that the AUC values at 1/3/5 years were greater than 0.6 for both risk score and N stage, suggesting a better prognostic ability for risk score and N stage (Supplementary Fig. 3C). Signaling pathways significantly associated with HNSCC prognosis were identified GSEA results indicated that top 5 signaling pathways significantly enriched for DEGs between HRG and LRG were ‘ribosome’, ‘huntingtons disease’, ‘dilated cardiomyopathy’, ‘hypertrophic cardiomyopathy hcm’, and ‘focal adhesion’ (Fig. 5 A). At the same time, GSVA results showed that ‘notch signaling pathway’, ‘fructose and mannose metabolism’, ‘pentose phosphate pathway’, ‘natural killer cell-mediated cytotoxicity’, ‘regulation of autophagy’ and other pathways were significantly different between HRG and LRG (Fig. 5 B). Moreover, univariate Cox regression analysis was conducted to explore the relationship between the top five pathways obtained from GSVA and survival in HNSCC patients. The results demonstrated that these five pathways were significantly related to the survival of HNSCC patients. Figure 4 The clinical utility of risk modeling. Univariate (A) and multivariate (B) Cox analysis based on risk scores and other clinical characteristics. (C) The nomogram for predicting the survival probability of HNSCC patients has two independent prognostic features. (D) The calibration plots of the nomogram for predicting OS probability for 1-, 3-, and 5-years. (E) ROC analysis of the nomogram. With the exception of the JAK/STAT signaling pathway, all other signaling pathways facilitated HNSCC progression (Fig. 5 C). Risk score was strongly associated with the immune microenvironment in HNSCC patients ESTIMATE algorithm results revealed considerable disparities in stromal scores, immune scores, and ESTIMATE scores between HRG and LRG, with significantly higher values observed in the LRG (Fig. 6 A). Meanwhile, the infiltration scores of 20 out of 28 immune cells were found to be considerably different between HRG and LRG. All immune cells except CD56bright natural killer cell had higher infiltration scores in LRG, including monocytes, neutrophils, and eosinophils (Fig. 6 B). The top five differential immune cells were plasmacytoid dendritic cell, T follicular helper cell, CD56bright natural killer cell, type 17 T helper cell, and mast cell ( Supplementary Table 2A) . Correlation analysis confirmed that CD56bright natural killer cell showed a significant positive correlation with the risk score, while other four immune cells were opposite (Fig. 6 C). Furthermore, significant differences in the survival of T follicular helper cell, Type 17 T helper cell, CD56bright natural killer cell, and mast cell between HRG and LRG revealed that all of these immune cells were associated with the survival of HNSCC ( Supplementary Fig. 4A-E ). Additionally, a noteworthy inverse correlation was observed between PRDX6 and plasmacytoid dendritic cell cells (cor=-0.36, P = 3.95×10 − 17 ), while a significant positive correlation was found between GSTO1 and CD56bright natural killer cells (cor = 0.37, P = 7.9×10 − 17 ) (Fig. 6 D, Supplementary Table 2B ). On the other hand, a significant disparity was observed in the expression levels of 10 immune checkpoints between the HRG and the LRG, namely CD276, CD27, CD40, CD70, ENTPD1, ICOS, SIGLEC15, TNFRSF18, TNFRSF9, TNFSF9. Among them, CD276, CD40, CD70, TNFRSF18, and TNFSF9 were significantly higher expressed in HRG, while the opposite was true for the remaining five immune checkpoints (Fig. 6 E). Also, the TIDE score, dysfunction score, and exclusion score for HNSCC samples between HRG and LRG were compared. The results demonstrated that there were no statistically considerable disparities in TIDE score between HRG and LRG. However, LRG exhibited significantly higher sensitivity to dysfunction scores, while HRG showed significantly greater sensitivity to exclusion scores (Fig. 6 F). The identification of drugs linked to risk score and the assessment of TMB Drug sensitivity analysis revealed that there was a considerable association between 15 drugs and risk scores, namely doramapimod_1042, GDC0810_1925, KU-55933_1030, NU7441_1038, pevonedistat_1529, GSK591_2110, rapamycin_1084, ZM447439_1050, AZD6738_1917, paclitaxel_1080, AZD7762_1022, docetaxel_1007, ABT737_1910, JQ1_2172, and AZD8186_1918. Importantly, doramapimod_1042 was considerably positively associated with the risk score (cor = 0.43, P < 2.2×10 − 16 ), while docetaxel_1007 had a considerable negative relationship with risk score (cor=-0.33, P = 1.4×10 − 13 ) (Fig. 7 A). In the TCGA-HNSCC dataset, somatic mutations of patients in HRG and LRG were analyzed. The results indicated that TP53 and TTN were the top two mutated genes in both HRG and LRG ( Supplementary Fig. 5A-B ). In addition, we observed a significant increase in TMB values in HRG ( Supplementary Fig. 5C ) and found a strong correlation between these values and survival outcomes among patients with HNSCC ( Supplementary Fig. 5D ). These results indicated that the involvement of TMB was significant in the progression of HNSCC. Prognostic genes expression changed with different stages of cell differentiation In the GSE140042 dataset, there were 21,489 cells and 33,538 genes prior to QC ( Supplementary Fig. 6A ), whereas post-QC analysis generated 11,631 cells and 33,538 genes ( Supplementary Fig. 6B ). Then, the “vst” method was applied to obtain the top 2,000 highly variable genes ( Supplementary Fig. 6C ). PCA results showed no significant outliers in the samples from GSE140042 dataset ( Supplementary Fig. 6D ). Following this, 20 PCs were identified for subsequent analysis ( Supplementary Fig. 6E ). Immediately after, 30 cell clusters were obtained by clustering analysis (Fig. 8 A), followed by demonstrating the expression of marker genes in these cell clusters (Fig. 8 B). Furthermore, six cell types were annotated, including T cells, CD4 + T cells, fibroblasts, CD8 + T cells, B cells, and epithelial cells (Fig. 8 C). Importantly, epithelial cells were selected as key cells due to higher expression of prognostic genes (Fig. 8 D). Nine subtypes were obtained by dimensionality reduction ( Supplementary Fig. 6F ) and clustering analysis (Fig. 8 E) of epithelial cells, followed by annotation to determine the specific cell types of these subtypes (Fig. 8 F). Pseudo-time analysis for key cells revealed that the cell subtypes of epithelial cells, urothelium cell, premeiotic germ cell and endoderm progenitor cell, were at state1, state4 and state5 of epithelial cell differentiation, respectively (Fig. 8 G). Then, 2,000 marker genes with significant changes were extracted, and the genes were clustered to three clusters. The heat map revealed the changes in expression of these three gene clusters (Fig. 8 H). Moreover, the expression of ADA and PRDX6 demonstrated a temporal pattern. As the pseudo-time changed, their expression first decreased and then increased. Conversely, GSTO1 displayed an opposing trend; its expression initially increased and subsequently decreased. In stark contrast, SMS showed no substantial change in its expression over the pseudo-time scale. (Fig. 8 I). Discussion HNSCC is a type of common malignant tumor, which seriously threatens the life and health of patients. As essential amino acids, BCAAs have attracted much attention in recent years in the field of tumor research, including their effects on tumor cell metabolism, signaling pathways, immune regulation, and TME 47 , 48 .The study combined RNA-seq and scRNA-seq to construct and validate an BCAA-MRGs risk model. The results showed that the model could accurately predict the prognosis of HNSCC patients, and the expression levels of prognostic genes were verified by RT-qPCR. These findings provide directions for the specific treatment of HNSCC patients. Previous studies have shown that BCAA metabolism affects different cellular processes, ranging from protein synthesis to epigenetics 49 . Disturbed BCAA metabolism in cancer may affect these processes and lead to disease progression, potentially serving as a marker for predicting disease status. Therefore, it is crucial to study the genes associated with BCAA metabolism in HNSCC. In this study, SMS, PRDX6, GSTO1 and ADA were identified as prognostic genes for risk modeling. The HNSCC samples were categorized into HRG and LRG based on the optimal threshold of the risk score, with LRG having a significantly higher overall survival rate than HRG. The SMS (spermine synthase) gene, a member of the spermidine/spermine synthase family, is located on chromosomes 1, 5, 6, and X 50 . Recent studies have shown that SMS is associated with the progression of a variety of cancers, including hepatocellular carcinoma 51 , breast cancer 52 , colorectal cancer 53 , and HNSCC 54 . Pan et al. found that SMS expression was elevated in HNSCC compared to normal tissues, and poor overall survival was associated with high SMS expression 54 . These findings are consistent with ours. Upregulation of SMS expression leads to increased intracellular levels of spermine, one of the polyamines, which promotes tumor cell invasion, proliferation and metastasis while inhibiting apoptosis. In tumor cells, the mTOR signaling pathway is both affected by polyamine levels and involved in the regulation of BCAA metabolism 55 , 56 . Polyamine may indirectly affect the expression of BCAA-MRGs and enzyme activities by regulating the activity of the mTOR signaling pathway, thus forming a complex metabolic regulatory network in tumor cells. PRDX6, a member of the antioxidant enzyme superfamily, plays an important role in cell differentiation, apoptosis, and redox-mediated signaling 57 . Data analysis showed that PRDX6 expression was down-regulated in HNSCC tumor tissues compared with normal tissues, and the expression trend of PRDX6 was found to be consistent in HNSCC tumor tissues by RT-qPCR, although there was no significant difference in the expression, which might be related to the heterogeneity of the samples. PRDX6 was found to be a poor prognostic factor for tongue squamous cell carcinoma, and its high expression was significantly correlated with shortened OS in patients 58 . In addition, PRDX6 expression was up-regulated in the subgroup with poor histopathological differentiation of HNSCC, and the high expression group showed reduced OS 59 , suggesting that this gene is an independent prognostic factor for HNSCC. This is consistent with our findings, where our univariate Cox regression analysis results showed that PRDX6 is a risk factor for HNSCC. It has been shown that in HNSCC, PRDX6 inhibits apoptosis of tumor cells by exerting oxidative effects 60 . Given the important role of PRDX6 in HNSCC, it is expected to be a potential therapeutic target. GSTO1 (Glutathione S-transferase omega 1) is a multifunctional enzyme with detoxification, redox regulation, and is involved in the regulation of signaling pathways in a variety of pathological diseases including cancer. GSTO1 expression is up-regulated in many cancers, such as bladder cancer 61 , non-small cell lung cancer 62 , and esophageal squamous cell carcinomas 63 , and is in connection with poor prognosis. In the present study, we also found that high expression of GSTO1 was associated with poor prognosis in HNSCC. GSTO1 plays an important role in the regulation of intracellular redox state. It was found that, on the one hand, the energy and metabolites generated by BCAA metabolism can regulate the level of reactive oxygen species (ROS) and affect cell proliferation. On the other hand, ROS can alter the supply and metabolism of BCAA by affecting BCAA metabolism-related enzymes and transporter proteins, which in turn affects the activity of the mTOR signaling pathway and promotes or inhibits tumor proliferation 56 , 64 . Accordingly, we hypothesize that GSTO1 may affect the metabolism of BCAA through regulating the level of intracellular ROS and influence the development of tumors. Adenosine deaminase (ADA) contains two isoenzymes, ADA1 and ADA2, which are mainly generated from adenosine triphosphate (ATP) through a series of enzymatic reactions, and adenosine is an important immunosuppressive signal 65 , 66 . ADA affects cell proliferation, differentiation, and apoptosis by regulating intracellular adenosine levels. Research has found that elevated BCAA levels alter ATP production, interfering with glycolysis, fatty acid oxidation, the tricarboxylic acid (TCA) cycle, and oxidative phosphorylation, leading to mitochondrial dysfunction 67 . Adenosine may be the key step linking ADA to BCAA metabolism. In summary, there is a close and complex interaction between BCAA metabolism and tumors, and this interaction plays an important role in tumor development by affecting BCAA metabolism-related enzymes and transport proteins. It is expected that further in-depth study of BCAA metabolism to provide new targets and strategies for the diagnosis and treatment of patients with HNSCC, and to bring hope for the improvement of the prognosis of patients with HNSCC. To further explore the pathogenic mechanism of HNSCC, we performed GSEA and GSVA pathway enrichment analysis and univariate Cox regression analysis of GSVA enrichment results, and found that “fructose and mannose metabolism”, “pentose phosphate pathway” and “natural killer cell-mediated cytotoxicity” as well as Wnt signaling pathway were upregulated and significantly connected with poor prognosis of HNSCC. It is well known that many cancers exhibit a strong glycolytic phenotype, which usually correlates with the aggressiveness and malignancy of the tumor 68 , 69 . In cancer, cytoplasmic glucose can also be utilized through the pentose phosphate pathway, which induces cell proliferation through the synthesis of nucleic acids, fatty acids, and amino acids 70 . Jonathan et al. found higher glycolysis scores in HNSCC by RNA-seq profiling of glycolytic tumors in a pan-cancer cohort 71 . This is consistent with the results of the present study that glycolysis may induce amino acid synthesis in HNSCC by interacting with the pentose phosphate pathway, thereby promoting cancer cell proliferation. Therefore, the development of tumor therapeutic strategies targeting this metabolic pathway will offer novel strategies for the treatment of HNSCC. In addition, this study found that natural killer cell-mediated cytotoxicity pathway was upregulated and associated with poor prognosis of HNSCC. While the conventional view is that natural killer cells inhibit tumor growth through their cytotoxic effects, recent studies have found that NK cell-mediated cytotoxicity may promote tumor progression under certain circumstance 72 . For example, NK cells can not only promote epithelial-mesenchymal transition (EMT) of tumor cells by secreting cytokines and chemokines to enhance tumor migration and invasion, but also inhibit anti-tumor immune responses by attracting immunosuppressive cells into the TME, such as regulatory T cells. To further reveal the TME in HNSCC patients, we performed a series of TME correlation analyses. In the TME, there is a competitive relationship between immune cells and tumor cells for BCAAs. Immune cells can absorb and utilize BCAAs, inhibiting the progress of the tumor. When tumor cells consume a large amount of BCAAs, the access of immune cells to BCAAs can be inhibited, resulting in immune evasion 73 . In our study, we found significant differences in the TME between HRG and LRG. Among them, plasmacytoid dendritic cells, mast cells, type 17 T helper cells, and T follicular helper cells were infiltrated to a higher extent in the LRG than in the HRG, while CD56 bright natural killer cells were infiltrated to a low extent. Wulff et al. found lower numbers of circulating immunoregulatory CD56 bright natural killer cells in the peripheral blood of patients with HNSCC, regardless of individual tumor stage or tumor type 74 . This coincides with the results of the present study. The high - uptake of tumor cells can lead to local changes in the concentration of BCAAs in the TME. BCAAs are important nutrients for NK cell metabolism. In the absence of BCAAs, the metabolism of NK cells will be disrupted. CD56 + natural killer (NK) cells are essential for innate antiviral and antitumor host defense. Based on their CD56 surface density expression, they can be categorized into dim and bright subpopulations. Dim NK cells are cytolytic and account for approximately 90% of NK cells, whereas bright NK cells are immunomodulatory 75 . Lower infiltration of CD56 bright natural killer cells leads to decreased immunomodulation. In addition, this study found that the prognostic gene GSTO1 was significantly and positively correlated with CD56 bright natural killer cells. GSTO1 has the potential to affect the balance of activating and inhibiting receptor expression on the surface of NK cells, thereby altering the functional status of NK cells and influencing the progression of HNSCC. The remarkable plasticity of Th17 cells, a subset of T cells that can produce IL-22, IL-21 IL-17A, IL-17F, and IFN-γ, is an important player in inflammatory and autoimmune diseases as well as cancer 76 , 77 . In a study on oral cancer, it was found that mice with oral cancer had a low Th17 phenotype and that the Th17 phenotype declined with cancer progression 78 . Mast cells can remodel the TM, which may change the accessibility of tumor cells and other cells to BCAAs by influencing angiogenesis and nutrient transport in the TME. Mast cells can directly exert cytotoxic effects and induce apoptosis of tumor cells, and recruit and activate immune cells, including NK cells and cytotoxic T lymphocytes, to the tumor site, thus exerting anti-tumor effects 79 . In summary, immune cells have a complex role in HNSCC, promoting both tumor progression and anti-tumor activity. The present study is expected to enhance the anti-tumor function of immune cells by modulating the BCAA-related metabolic pathways to bring better therapeutic outcomes for HNSCC patients. In order to find potential drugs for the treatment of HNSCC patients, this study analyzed the drug sensitivity of a specific subgroup of HRG and found that they were sensitive to 15 drugs, including doramapimod, rapamycin, paclitaxel, and docetaxel. Doramapimod is a small molecule inhibitor of p38 mitogen-activated protein kinase (MAPK). The p38 MAPK signaling pathway plays a key role in tumor-related inflammation and immune regulation. Studies have found increased sensitivity to doramapimod chemotherapy regimens in esophageal squamous cell carcinoma 80 . Docetaxel belongs to the paclitaxel class of drugs, and its main mechanism of action is to promote the assembly of microtubule proteins into stable microtubules and inhibit microtubule depolymerization, which stops cell mitosis in the G2-M phase, thus inhibiting the proliferation of tumor cells. Docetaxel-platinum-cetuximab treatment showed promising results in a phase 2 trial of first-line recurrent or metastatic HNSCC 81 . Paclitaxel in combination with cisplatin or carboplatin is one of the classic chemotherapeutic regimens in the treatment of HNSCC. Paclitaxel acts on the microtubule system and serves to kill tumor cells 82 . In the future, with in-depth research on the pathogenesis of HNSCC and advances in drug development, chemotherapy treatment regimens are expected to be continuously optimized, bringing better therapeutic effects and quality of life to HNSCC patients. In order to reveal the detailed heterogeneity of prognostic genes in tumor cells and to compare the expression of prognostic genes in various cell clusters, we performed scRNA-seq analysis, and found that prognostic genes were expressed in CD4 + T cells, fibroblasts, and CD8 + T cells, especially in epithelial cells. It was found that in HNSCC, epithelial cells could interact with immune cells and fibroblasts to co-construct a microenvironment that promotes tumor growth and supports tumor progression 83 , 84 . Dysfunctional epithelial cells can enhance the invasiveness of migrating HNSCC tumor cells 85 , 86 . Different subpopulations of epithelial cells were further analyzed, and it was found that prognostic genes were differently expressed in different developmental states, and the expression levels of two genes, ADA and PRDX6, decreased and then increased with pseudo time, while GSTO1 increased and then decreased, and SMS showed an overall unchanged trend. In this study, we explored the role of BCAAs in HNSCC by combining RNA-seq and scRNA-seq data, and identified four key genes related to BCAA metabolism, based on which we constructed, for the first time, a reliable model that can predict the prognosis of HNSCC. In addition, we found that the key genes may be associated with the sensitivity of HNSCC patients to chemotherapeutic agents and immune checkpoint inhibitors. Moreover, we also found that in epithelial cells, the expression levels of key genes vary with different stages of cell differentiation. And the expression levels of the key genes were further validated experimentally. The identification of these key genes provides important clinical implications for the diagnosis and treatment of HNSCC. However, our study is inevitably subject to certain limitations. First, the main data for our study were obtained from TCGA and GEO databases. Second, the number of tumor samples we used to validate prognostic genes was small. In the future, we will continue to further explore the mechanism of branched-chain amino acid metabolism-related prognostic genes in HNSCC in various aspects of clinical and laboratory studies. Conclusion In summary, this study developed a reliable prognostic model based on BCAAs. And initially but relatively comprehensively explored the potential mechanisms by which these genes influence patient prognosis. This provides a new reference for the mechanism of action of BCAA-MRGs in HNSCC. Declarations Acknowledgements Data supporting the results of this study are publicly available from the TCGA databases (TCGA-HNSCC, https://gdc.cancer.gov/) and GEO databases (GSE65858 and GSE140042, https://www.ncbi.nlm. nih.gov/geo/). The experimental data will be made available from the corresponding author upon reasonable request.We thank all the individuals who participated in this study. We also thank the GEO and TCGA data platform. In Addition, we sincerely thank the Department of Otorhinolaryngology and Head and Neck Surgery, Xinjiang Uygur Autonomous Region People's Hospital for providing the study samples of squamous cell carcinoma of the head and neck. Data availability statement Data supporting the results of this study are publicly available from the TCGA databases (TCGA-HNSCC, https://gdc.cancer.gov/) and GEO databases (GSE65858 and GSE140042, https://www.ncbi.nlm. nih.gov/geo/). The experimental data will be made available from the corresponding author upon reasonable request. Author contributions Conceptualization, D.L. and A.A.; methodology, D.L. and Y.L.; resources, D.L. and A.A.; writing—original draft preparation, D.L. and T.M.; writing—review and editing, D.L., Y.L. and A.A.; visualization, Y.L. and T.M.; supervision, D.L. and A.A.; project administration, A.A.; funding acquisition, D.L. and A.A. All authors have read and agreed to the published version of the manuscript. Funding This work was supported by the Natural Science Foundation of Xinjiang Uygur Autonomous Region (2021D01C188) and the Science Foundation of 215 Hospital of Shaanxi Nuclear Industry (215KYJJ-202324). Ethical approval statement The data for this study were obtained from the open databases TCGA and GEO. Cancer tissues and paracancerous tissues are obtained from patients with Head and neck squamous cell carcinoma in the Xinjiang Uygur Autonomous Region People's Hospital. All experimental protocols were approved by the Research Ethics Committee of Xinjiang Uygur Autonomous Region People's Hospital. Declaration of Interest Statement The authors declare that they have no competing interests in this work. References Bray, F. et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin. 74 , 229–263. 10.3322/caac.21834 (2024). Zhao, X., Chen, H., Qiu, Y. & Cui, L. FAM64A promotes HNSCC tumorigenesis by mediating transcriptional autoregulation of FOXM1. Int. J. Oral Sci. 14 10.1038/s41368-022-00174-4 (2022). Nishimura, A. et al. Incurable locoregional disease is a strong poor prognostic factor in recurrent or metastatic squamous cell carcinoma of the head and neck. Int. J. Clin. Oncol. 26 , 1822–1830. 10.1007/s10147-021-01965-1 (2021). Xu, B. et al. Association of dietary intake of branched-chain amino acids with long-term risks of CVD, cancer and all-cause mortality. Public Health. Nutr. 25 , 1–11. 10.1017/s1368980021004948 (2021). Nouri-Majd, S., Salari-Moghaddam, A., Benisi-Kohansal, S., Azadbakht, L. & Esmaillzadeh, A. Dietary intake of branched-chain amino acids in relation to the risk of breast cancer. Breast cancer (Tokyo Japan) . 29 , 993–1000. 10.1007/s12282-022-01379-5 (2022). Sciacovelli, M. et al. Dynamic partitioning of branched-chain amino acids-derived nitrogen supports renal cancer progression. Nat. Commun. 13 , 7830. 10.1038/s41467-022-35036-4 (2022). Kim, S. K. & Cho, S. W. The Evasion Mechanisms of Cancer Immunity and Drug Intervention in the Tumor Microenvironment. Front. Pharmacol. 13 , 868695. 10.3389/fphar.2022.868695 (2022). Aktar, N. et al. Understanding of Immune Escape Mechanisms and Advances in Cancer Immunotherapy. Journal of oncology 8901326, (2022). 10.1155/2022/8901326 (2022). DePeaux, K. & Delgoffe, G. M. Metabolic barriers to cancer immunotherapy. Nat. Rev. Immunol. 21 , 785–797. 10.1038/s41577-021-00541-y (2021). Kristian, Y. Y. et al. Correlation between branched-chain amino acids intake and total lymphocyte count in head and neck cancer patients: a cross-sectional study. BMC Nutr. 9 , 86. 10.1186/s40795-023-00746-5 (2023). Dauphinais, C. & Waithe, W. I. PHA stimulation of human lymphocytes during amino acid deprivation. Protein, RNA and DNA synthesis. J. Cell. Physiol. 91 , 357–367. 10.1002/jcp.1040910305 (1977). Chuang, J. C., Yu, C. L. & Wang, S. R. Modulation of human lymphocyte proliferation by amino acids. Clin. Exp. Immunol. 81 , 173–176. 10.1111/j.1365-2249.1990.tb05310.x (1990). Wetzel, T. J., Erfan, S. C. & Ananieva, E. A. The emerging role of the branched chain aminotransferases, BCATc and BCATm, for anti-tumor T-cell immunity. Immunometabolism (Cobham Surrey) . 5 , e00014. 10.1097/in9.0000000000000014 (2023). Sinclair, L. V. et al. Control of amino-acid transport by antigen receptors coordinates the metabolic reprogramming essential for T cell differentiation. Nat. Immunol. 14 , 500–508. 10.1038/ni.2556 (2013). Waickman, A. T. & Powell, J. D. mTOR, metabolism, and the regulation of T-cell differentiation and function. Immunol. Rev. 249 , 43–58. 10.1111/j.1600-065X.2012.01152.x (2012). Pollizzi, K. N. & Powell, J. D. Regulation of T cells by mTOR: the known knowns and the known unknowns. Trends Immunol. 36 , 13–20. 10.1016/j.it.2014.11.005 (2015). Wang, W. & Zou, W. Amino Acids and Their Transporters in T Cell Immunity and Cancer Therapy. Mol. Cell . 80 , 384–395. 10.1016/j.molcel.2020.09.006 (2020). Blair, M. C., Neinast, M. D. & Arany, Z. Whole-body metabolic fate of branched-chain amino acids. Biochem. J. 478 , 765–776. 10.1042/bcj20200686 (2021). Rossi, M. et al. Dietary intake of branched-chain amino acids and colorectal cancer risk. Br. J. Nutr. 126 , 22–27. 10.1017/s0007114520003724 (2021). Yu, Z., Qiu, B., Zhou, H., Li, L. & Niu, T. Characterization and application of a lactate and branched chain amino acid metabolism related gene signature in a prognosis risk model for multiple myeloma. Cancer Cell Int. 23 , 169. 10.1186/s12935-023-03007-4 (2023). Zheng, J. et al. Integrated single-cell and bulk characterization of branched chain amino acid metabolism-related key gene BCAT1 and association with prognosis and immunogenicity of clear cell renal cell carcinoma. Aging 16 , 2715–2735. 10.18632/aging.205506 (2024). Li, J. Y. et al. GEO data mining and TCGA analysis reveal altered branched chain amino acid metabolism in pancreatic cancer patients. Aging 13 , 11907–11918. 10.18632/aging.202892 (2021). Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome biology 15, 550, (2014). 10.1186/s13059-014-0550-8 Gustavsson, E. K., Zhang, D., Reynolds, R. H., Garcia-Ruiz, S. & Ryten, M. ggtranscript: an R package for the visualization and interpretation of transcript isoforms using ggplot2. Bioinf. (Oxford England) . 38 , 3844–3846. 10.1093/bioinformatics/btac409 (2022). Gu, Z., Eils, R. & Schlesner, M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. Bioinf. (Oxford England) . 32 , 2847–2849. 10.1093/bioinformatics/btw313 (2016). Hänzelmann, S., Castelo, R. & Guinney, J. GSVA: gene set variation analysis for microarray and RNA-seq data. BMC Bioinform. 14 10.1186/1471-2105-14-7 (2013). Liu, T. T. et al. Identification of CDK2-Related Immune Forecast Model and ceRNA in Lung Adenocarcinoma, a Pan-Cancer Analysis. Front. cell. Dev. biology . 9 , 682002. 10.3389/fcell.2021.682002 (2021). Langfelder, P. & Horvath, S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinform. 9 10.1186/1471-2105-9-559 (2008). Chen, H. & Boutros, P. C. VennDiagram: a package for the generation of highly-customizable Venn and Euler diagrams in R. BMC Bioinform. 12 10.1186/1471-2105-12-35 (2011). Wu, T. et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innov. (Cambridge (Mass)) . 2 , 100141. 10.1016/j.xinn.2021.100141 (2021). Shannon, P. et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 13 , 2498–2504. 10.1101/gr.1239303 (2003). Lei, J. et al. Clinicopathological characteristics of pheochromocytoma/paraganglioma and screening of prognostic markers. J. Surg. Oncol. 128 , 510–518. 10.1002/jso.27358 (2023). Fang, Y., Huang, S., Han, L., Wang, S. & Xiong, B. Comprehensive Analysis of Peritoneal Metastasis Sequencing Data to Identify LINC00924 as a Prognostic Biomarker in Gastric Cancer. Cancer Manage. Res. 13 , 5599–5611. 10.2147/cmar.S318704 (2021). Engebretsen, S. & Bohlin, J. Statistical predictions with glmnet. Clin. epigenetics . 11 , 123. 10.1186/s13148-019-0730-1 (2019). Heagerty, P. J., Lumley, T. & Pepe, M. S. Time-dependent ROC curves for censored survival data and a diagnostic marker. Biometrics 56 , 337–344. 10.1111/j.0006-341x.2000.00337.x (2000). Xu, J. et al. A nomogram for predicting prognosis of patients with cervical cerclage. Heliyon 9 , e21147. 10.1016/j.heliyon.2023.e21147 (2023). Mak, T. K. et al. The cancer-associated fibroblast-related signature predicts prognosis and indicates immune microenvironment infiltration in gastric cancer. Front. Immunol. 13 , 951214. 10.3389/fimmu.2022.951214 (2022). Yang, Y. et al. A composite score based on immune-related gene prognostic index and m(6)A risk score of head and neck squamous cell carcinoma. Front. Genet. 14 10.3389/fgene.2023.1061569 (2023). Jędroszka, D., Orzechowska, M., Hamouz, R., Górniak, K. & Bednarek, A. K. Markers of epithelial-to-mesenchymal transition reflect tumor biology according to patient age and Gleason score in prostate cancer. PloS one . 12 , e0188842. 10.1371/journal.pone.0188842 (2017). Robles-Jimenez, L. E. et al. Worldwide Traceability of Antibiotic Residues from Livestock in Wastewater and Soil: A Systematic Review. Animals: open. access. J. MDPI . 12 10.3390/ani12010060 (2021). Cheng, J., Chen, F. & Cheng, Y. Construction and Evaluation of a Risk Score Model for Lymph Node Metastasis-Associated Circadian Clock Genes in Esophageal Squamous Carcinoma. Cells 11, (2022). 10.3390/cells11213432 Mayakonda, A., Lin, D. C., Assenov, Y., Plass, C. & Koeffler, H. P. Maftools: efficient and comprehensive analysis of somatic variants in cancer. Genome Res. 28 , 1747–1756. 10.1101/gr.239244.118 (2018). Maeser, D., Gruener, R. F. & Huang, R. S. oncoPredict: an R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data. Brief. Bioinform. 22 10.1093/bib/bbab260 (2021). Hao, Y. et al. Integrated analysis of multimodal single-cell data. Cell 184 , 3573–3587e3529. 10.1016/j.cell.2021.04.048 (2021). Hao, X. et al. Inhibition of APOC1 promotes the transformation of M2 into M1 macrophages via the ferroptosis pathway and enhances anti-PD1 immunotherapy in hepatocellular carcinoma based on single-cell RNA sequencing. Redox Biol. 56 , 102463. 10.1016/j.redox.2022.102463 (2022). Qiu, X. et al. Reversed graph embedding resolves complex single-cell trajectories. Nat. Methods . 14 , 979–982. 10.1038/nmeth.4402 (2017). Xu, E., Ji, B., Jin, K. & Chen, Y. Branched-chain amino acids catabolism and cancer progression: focus on therapeutic interventions. Front. Oncol. 13 , 1220638. 10.3389/fonc.2023.1220638 (2023). Bonvini, A., Coqueiro, A. Y., Tirapegui, J., Calder, P. C. & Rogero, M. M. Immunomodulatory role of branched-chain amino acids. Nutr. Rev. 76 , 840–856. 10.1093/nutrit/nuy037 (2018). Sivanand, S. & Vander Heiden, M. G. Emerging Roles for Branched-Chain Amino Acid Metabolism in Cancer. Cancer cell. 37 , 147–156. 10.1016/j.ccell.2019.12.011 (2020). Pegg, A. E. & Michael, A. J. Spermine synthase. Cell. Mol. Life Sci. 67 , 113–121. 10.1007/s00018-009-0165-5 (2010). Xiang, L., Piao, L., Wang, D. & Qi, L. F. Overexpression of SMS in the tumor microenvironment is associated with immunosuppression in hepatocellular carcinoma. Front. Immunol. 13 , 974241. 10.3389/fimmu.2022.974241 (2022). Fahrmann, J. F. et al. Association Between Plasma Diacetylspermine and Tumor Spermine Synthase With Outcome in Triple-Negative Breast Cancer. J. Natl Cancer Inst. 112 , 607–616. 10.1093/jnci/djz182 (2020). Guo, Y. et al. Spermine synthase and MYC cooperate to maintain colorectal cancer cell survival by repressing Bim expression. Nat. Commun. 11 , 3243. 10.1038/s41467-020-17067-x (2020). Pan, X., Xue, L. & Sun, Y. Spermine synthase (SMS) serves as a prognostic biomarker in head and neck squamous cell carcinoma: a bioinformatics analysis. Annals translational Med. 10 , 1213. 10.21037/atm-22-5014 (2022). Novita Sari, I. et al. Metabolism and function of polyamines in cancer progression. Cancer Lett. 519 , 91–104. 10.1016/j.canlet.2021.06.020 (2021). Jung, M. K., Okekunle, A. P., Lee, J. E., Sung, M. K. & Lim, Y. J. Role of Branched-chain Amino Acid Metabolism in Tumor Development and Progression. J. cancer Prev. 26 , 237–243. 10.15430/jcp.2021.26.4.237 (2021). Ding, C., Fan, X. & Wu, G. Peroxiredoxin 1 - an antioxidant enzyme in cancer. J. Cell. Mol. Med. 21 , 193–202. 10.1111/jcmm.12955 (2017). Yin, K. et al. Using weighted gene co-expression network analysis to identify key modules and hub genes in tongue squamous cell carcinoma. Medicine 98 , e17100. 10.1097/md.0000000000017100 (2019). Xu, L. et al. A Novel Ferroptosis-Related Gene Signature to Predict Prognosis in Patients with Head and Neck Squamous Cell Carcinoma. Disease markers 5759927, (2021). 10.1155/2021/5759927 (2021). Leopoldino, A. M. et al. SET protein accumulates in HNSCC and contributes to cell survival: antioxidant defense, Akt phosphorylation and AVOs acidification. Oral Oncol. 48 , 1106–1113. 10.1016/j.oraloncology.2012.05.014 (2012). Djukic, T. et al. Upregulated glutathione transferase omega-1 correlates with progression of urinary bladder carcinoma. Redox report: Commun. free radical Res. 22 , 486–492. 10.1080/13510002.2017.1299909 (2017). Wang, K., Zhang, F. L. & Jia, W. Glutathione S–transferase ω 1 promotes the proliferation, migration and invasion, and inhibits the apoptosis of non–small cell lung cancer cells, via the JAK/STAT3 signaling pathway. Mol. Med. Rep. 23 10.3892/mmr.2020.11709 (2021). Li, Y. et al. Identification of glutathione S-transferase omega 1 (GSTO1) protein as a novel tumor-associated antigen and its autoantibody in human esophageal squamous cell carcinoma. Tumour biology: J. Int. Soc. Oncodevelopmental Biology Med. 35 , 10871–10877. 10.1007/s13277-014-2394-y (2014). Zhenyukh, O. et al. Branched-chain amino acids promote endothelial dysfunction through increased reactive oxygen species generation and inflammation. J. Cell. Mol. Med. 22 , 4948–4962. 10.1111/jcmm.13759 (2018). Zavialov, A. V. & Engström, A. Human ADA2 belongs to a new family of growth factors with adenosine deaminase activity. Biochem. J. 391 , 51–57. 10.1042/bj20050683 (2005). Kaljas, Y. et al. Human adenosine deaminases ADA1 and ADA2 bind to different subsets of immune cells. Cell. Mol. Life Sci. 74 , 555–570. 10.1007/s00018-016-2357-0 (2017). Ye, Z., Wang, S., Zhang, C. & Zhao, Y. Coordinated Modulation of Energy Metabolism and Inflammation by Branched-Chain Amino Acids and Fatty Acids. Front. Endocrinol. 11 , 617. 10.3389/fendo.2020.00617 (2020). Feng, J. et al. Emerging roles and the regulation of aerobic glycolysis in hepatocellular carcinoma. J. experimental Clin. cancer research: CR . 39 , 126. 10.1186/s13046-020-01629-4 (2020). Zhang, Z. et al. HOXA1 promotes aerobic glycolysis and cancer progression in cervical cancer. Cell. Signal. 109 , 110747. 10.1016/j.cellsig.2023.110747 (2023). Alfarouk, K. O. et al. The Pentose Phosphate Pathway Dynamics in Cancer and Its Dependency on Intracellular pH. Metabolites 10, (2020). 10.3390/metabo10070285 Chacon-Barahona, J. A., MacKeigan, J. P. & Lanning, N. J. Unique Metabolic Contexts Sensitize Cancer Cells and Discriminate between Glycolytic Tumor Types. Cancers 15 10.3390/cancers15041158 (2023). Xu, X. et al. Association of Germline Variants in Natural Killer Cells With Tumor Immune Microenvironment Subtypes, Tumor-Infiltrating Lymphocytes, Immunotherapy Response, Clinical Outcomes, and Cancer Risk. JAMA Netw. open. 2 , e199292. 10.1001/jamanetworkopen.2019.9292 (2019). Wetzel, T. J., Erfan, S. C., Figueroa, L. D., Wheeler, L. M. & Ananieva, E. A. Crosstalk between arginine, glutamine, and the branched chain amino acid metabolism in the tumor microenvironment. Front. Oncol. 13 , 1186539. 10.3389/fonc.2023.1186539 (2023). Wulff, S., Pries, R., Börngen, K., Trenkle, T. & Wollenberg, B. Decreased levels of circulating regulatory NK cells in patients with head and neck cancer throughout all tumor stages. Anticancer Res. 29 , 3053–3057 (2009). Cooper, M. A. et al. Human natural killer cells: a unique innate immunoregulatory role for the CD56(bright) subset. Blood 97 , 3146–3151. 10.1182/blood.v97.10.3146 (2001). Loos, J. et al. Functional characteristics of Th1, Th17, and ex-Th17 cells in EAE revealed by intravital two-photon microscopy. J. Neuroinflamm. 17 10.1186/s12974-020-02021-x (2020). Feng, S., Chen, X. M., Wang, J. F. & Xu, X. Q. Th17 cells associated cytokines and cancer. Eur. Rev. Med. Pharmacol. Sci. 20 , 4032–4040 (2016). Young, M. R., Levingston, C. A. & Johnson, S. D. Treatment to sustain a Th17-type phenotype to prevent skewing toward Treg and to limit premalignant lesion progression to cancer. Int. J. Cancer . 138 , 2487–2498. 10.1002/ijc.29989 (2016). Cai, Z., Tang, B., Chen, L. & Lei, W. Mast cell marker gene signature in head and neck squamous cell carcinoma. BMC cancer . 22 , 577. 10.1186/s12885-022-09673-3 (2022). Wang, Z. et al. Genetic and molecular characterization of metabolic pathway-based clusters in esophageal squamous cell carcinoma. Sci. Rep. 14 , 6200. 10.1038/s41598-024-56391-w (2024). Guigay, J. et al. Cetuximab, docetaxel, and cisplatin versus platinum, fluorouracil, and cetuximab as first-line treatment in patients with recurrent or metastatic head and neck squamous-cell carcinoma (GORTEC 2014-01 TPExtreme): a multicentre, open-label, randomised, phase 2 trial. Lancet Oncol. 22 , 463–475. 10.1016/s1470-2045(20)30755-5 (2021). Forastiere, A. A. Use of paclitaxel (Taxol) in squamous cell carcinoma of the head and neck. Semin. Oncol. 20 , 56–60 (1993). Liu, S., Lian, M., Han, B., Fang, J. & Wang, Z. Single-cell integrated transcriptomics reveals the role of keratinocytes in head and neck squamous cell carcinoma. J. Appl. Genet. 65 , 727–745. 10.1007/s13353-024-00842-7 (2024). Sun, M. et al. Cuproptosis-related lncRNA JPX regulates malignant cell behavior and epithelial-immune interaction in head and neck squamous cell carcinoma via miR-193b-3p/PLAU axis. Int. J. Oral Sci. 16 , 63. 10.1038/s41368-024-00314-y (2024). Lai, Y. J. et al. CSC-3436 inhibits TWIST-induced epithelial-mesenchymal transition via the suppression of Twist/Bmi1/Akt pathway in head and neck squamous cell carcinoma. J. Cell. Physiol. 234 , 9118–9129. 10.1002/jcp.27589 (2019). Zhao, M. et al. Targeting of EZH2 inhibits epithelial–mesenchymal transition in head and neck squamous cell carcinoma via regulating the STAT3/VEGFR2 axis. 55 , 1165–1175, (2019). 10.3892/ijo.2019.4880 Additional Declarations No competing interests reported. Supplementary Files SupplementaryTable1.docx SupplementaryTable2.xlsx SupplementaryFigure1.jpg SupplementaryFigure2.jpg SupplementaryFigure3.jpg SupplementaryFigure4.jpg SupplementaryFigure5.jpg SupplementaryFigure6.jpg 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-6030799","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":458378076,"identity":"39c26abb-bf30-4b67-a94e-efc621880f00","order_by":0,"name":"Dandan Lu","email":"","orcid":"","institution":"Shaanxi Nuclear Industry 215 Hospital","correspondingAuthor":false,"prefix":"","firstName":"Dandan","middleName":"","lastName":"Lu","suffix":""},{"id":458378077,"identity":"2365f40e-f6df-46e5-b3fc-7bbb09c50671","order_by":1,"name":"Yongjun Liang","email":"","orcid":"","institution":"Shaanxi Nuclear Industry 215 Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yongjun","middleName":"","lastName":"Liang","suffix":""},{"id":458378078,"identity":"d1443d23-c770-4339-8ba1-86f9b34de67c","order_by":2,"name":"Tao Mo","email":"","orcid":"","institution":"Guangzhou Otolaryngology-Head and Neck Surgery Hospital","correspondingAuthor":false,"prefix":"","firstName":"Tao","middleName":"","lastName":"Mo","suffix":""},{"id":458378079,"identity":"14b54d7a-0020-41b7-bd9b-9547c49e5140","order_by":3,"name":"Abdeyrim Arikin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAs0lEQVRIiWNgGAWjYFACxmaGBAYGOQZmUrUYk6IFojaxgWj1/LMPNxs83HE4fX4778EPDDU20QS1SJxLbE5IPHM4d8NhvmQJhmNpuYStO8PYfCCxDaiFmcdAgrHhMGEt8lAt6fLNPMY/iNJiANSSANSSwHCYx4w4WwyBWgwS29INNwC1WCQQ4xe5M+yPJX+2WcvL958xvvGhxoYI70NAM4RKIFI5CNSRoHYUjIJRMApGHAAAtFQ9Mgb/64cAAAAASUVORK5CYII=","orcid":"","institution":"Guangzhou Otolaryngology-Head and Neck Surgery Hospital","correspondingAuthor":true,"prefix":"","firstName":"Abdeyrim","middleName":"","lastName":"Arikin","suffix":""}],"badges":[],"createdAt":"2025-02-14 12:53:52","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6030799/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6030799/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":85386667,"identity":"d54366e4-5d6b-4194-8596-ff89f191956f","added_by":"auto","created_at":"2025-06-25 09:53:48","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":271419,"visible":true,"origin":"","legend":"\u003cp\u003eIdentification of differential BCAA-MRGs in HNSCC. (A, B) Volcano and heatmap demonstrated the expression of differential genes between HNSCC patients and normal tissues. (C) Kaplan Meier analysis of the high-score group and the low-score group. Sample clustering map (D), soft threshold screening map (E), gene clustering dendrogram (F), and module-score correlation heatmap (G), and scatterplot of GS vs. MM in key modules (H) in WGCNA analysis. *\u003cem\u003ep\u003c/em\u003e\u0026lt; 0.05, **\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01, ***\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/34b981cac030c0a540106158.jpg"},{"id":85385430,"identity":"3430ddc0-2e4f-4dff-9751-b1ed1479d57c","added_by":"auto","created_at":"2025-06-25 09:45:48","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":190960,"visible":true,"origin":"","legend":"\u003cp\u003eScreening of candidate genes and functional enrichment. (A) Venn diagram showing the intersection of DEGs and BCAA_MRGs. (B, C) GO and KEGG analyses of differentially expressed BCAA-MRGs.\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/890952dbd130cdc67bd67c16.jpg"},{"id":85385429,"identity":"37a89f38-c235-479b-9a2b-e448570187b8","added_by":"auto","created_at":"2025-06-25 09:45:48","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":235195,"visible":true,"origin":"","legend":"\u003cp\u003eConstruction and validation of risk models. (A) Identification of the prognosis-related differentially expressed BCAA-MRGs by univariate Cox regression analysis. (B) LASSO Cox regression analysis of 4 prognosis-related differentially expressed BCAA-MRGs. (C) Expression levels of four BCAA-MRGs associated with prognosis between HNSCC tumor tissues and normal tissues. (D-G) The expression levels of four prognostic genes by RT-qPCR. (H-K) The Kaplan Meier analysis and ROC curve analysis in training and validation sets. *\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05, **\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01, ***\u003cem\u003ep\u003c/em\u003e\u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/628a08b369d19837034ea35c.jpg"},{"id":85383782,"identity":"2280f50f-1d5a-4a8c-acc2-a841ca2729ac","added_by":"auto","created_at":"2025-06-25 09:37:48","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":175656,"visible":true,"origin":"","legend":"\u003cp\u003eThe clinical utility of risk modeling. Univariate (A) and multivariate (B) Cox analysis based on risk scores and other clinical characteristics. (C) The nomogram for predicting the survival probability of HNSCC patients has two independent prognostic features. (D) The calibration plots of the nomogram for predicting OS probability for 1-, 3-, and 5-years. (E) ROC analysis of the nomogram.\u003c/p\u003e","description":"","filename":"Picture4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/3b6f930ee67ac214a285e1d6.jpg"},{"id":85383788,"identity":"ddd27afd-2c18-4222-8e24-6e7bea252a83","added_by":"auto","created_at":"2025-06-25 09:37:48","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":129344,"visible":true,"origin":"","legend":"\u003cp\u003eFunctional enrichment analyses between low‐and high-risk groups. GSEA (A) and GSVA enrichment analysis (B) between high-and low-risk groups. (C) Relevant pathways enriched by univariate Cox analysis were associated with survival in HNSCC patients.\u003c/p\u003e","description":"","filename":"Picture5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/6103d43b45b45ef5d7faf85e.jpg"},{"id":85383798,"identity":"d9dfd623-6248-494e-931f-8c5ff53e7236","added_by":"auto","created_at":"2025-06-25 09:37:49","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":279065,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted evaluation of immune microenvironment characteristics in HNSCC. (A) The difference of the ESTIMATE score, Immune score and Stromal score between the low-and high-risk groups. (B) The differences in the proportions of 28 immune cells between the low- and high-risk groups. (C) The correlation between the signature and infiltration abundances of four immune cell types. (D) The correlation analysis of 20 differential immune cells with 4 prognostic genes. (E) The expression of immune checkpoints between high- and low-risk groups. (F) The analysis of TIDE scores between high- and low-risk groups. *\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05, **\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.01, ***\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.001.\u003c/p\u003e","description":"","filename":"Picture6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/2f1cc684020b90a6e4b4f40b.jpg"},{"id":85383783,"identity":"8d831eff-f240-4763-b52b-1b76fa53afb5","added_by":"auto","created_at":"2025-06-25 09:37:48","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":269123,"visible":true,"origin":"","legend":"\u003cp\u003eDrug sensitivity analysis between different risk groups (A).\u003c/p\u003e","description":"","filename":"Picture7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/eb90f338a8cbd6f249d56b25.jpg"},{"id":85385438,"identity":"4c8eabb7-87bc-4bae-806e-d6ce271f6a58","added_by":"auto","created_at":"2025-06-25 09:45:49","extension":"jpg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":70578,"visible":true,"origin":"","legend":"\u003cp\u003eProcessing and analysis of single cell sequencing data of HNSCC. (A) Cell clustering analysis. (B) The mark gene expression bubble chart. (C) Cell Annotation Analysis. (D) The differential expression of four BCAA metabolism-related prognostic genes in different immune cell populations. (E) tSNE algorithm divides cells into 9 clusters. (F) The annotation analysis of key cells. (G) Pseudotime and trajectory analysis. (H) The heatmap of modular gene expression in HNSCC. (I) The expression levels of four BCAA metabolism-related prognostic genes in different developmental states of cells.\u003c/p\u003e","description":"","filename":"Picture8.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/06cf2a6aa860ee836cfeeb2f.jpg"},{"id":86332358,"identity":"09e9925b-eb0c-437d-920c-edb790d5e60e","added_by":"auto","created_at":"2025-07-09 12:32:41","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3222713,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/c77af754-2917-4d1f-8d33-a6b74fa5dec0.pdf"},{"id":85385427,"identity":"1e4c1154-3f3b-45d2-9624-d5a2c35a0918","added_by":"auto","created_at":"2025-06-25 09:45:48","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":16169,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTable1.docx","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/2ec2f4dcc64879e226e5a893.docx"},{"id":85383776,"identity":"f20f07ea-9e30-47de-8eed-e2c50b2208d7","added_by":"auto","created_at":"2025-06-25 09:37:48","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":15581,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTable2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/27360cbb62cdd9eb43d35c18.xlsx"},{"id":85383777,"identity":"e7f36407-3f07-44ef-87d7-064e8e32e81c","added_by":"auto","created_at":"2025-06-25 09:37:48","extension":"jpg","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":770412,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/3f07fa66d9bf76a4bf1fcc25.jpg"},{"id":85385431,"identity":"1d3cc33a-406d-4446-be2e-8b09f8db4059","added_by":"auto","created_at":"2025-06-25 09:45:48","extension":"jpg","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":816883,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/047b964006087b6e397d1bf6.jpg"},{"id":85383784,"identity":"c0b3d860-8c63-4f20-88ac-6bc0fb0bcd1b","added_by":"auto","created_at":"2025-06-25 09:37:48","extension":"jpg","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":712214,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/1f446fed0c698450f45e49a4.jpg"},{"id":85383790,"identity":"b55358fd-70b2-44d3-89d5-7c07c7c96fcd","added_by":"auto","created_at":"2025-06-25 09:37:48","extension":"jpg","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":1095648,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/8113d8c7d9ebe8374bcc1f82.jpg"},{"id":85383792,"identity":"d07a8cd6-d994-4ae1-973b-21d0b04b9ec8","added_by":"auto","created_at":"2025-06-25 09:37:48","extension":"jpg","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":904938,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/ca0cb2a07e1f7914ab6ead4d.jpg"},{"id":85385436,"identity":"fc9287be-c1f1-4b59-9f7e-21ebee3df6cb","added_by":"auto","created_at":"2025-06-25 09:45:49","extension":"jpg","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":945857,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6030799/v1/038f46729fd05c64ec7cb399.jpg"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integration RNA bulk and single cell RNA sequencing to explore the change of BCAA metabolism-related immune microenvironment and construct prognostic signature in HNSCC","fulltext":[{"header":"Introduction","content":"\u003cp\u003eHead and neck tumors have the seventh highest incidence rate of all cancers in the 2022 National Cancer Statistics\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. HNSCC is the most common pathologic type of head and neck tumors. Despite impressive advances in surgery, radiation, chemotherapy, targeted therapies, and especially immunotherapy in recent years, the prognosis for HNSCC is dishearteningly poor, as the 5-year survival rate hovers at merely around 50%\u003csup\u003e2\u003c/sup\u003e. In addition, due to the heterogeneity of HNSCC, histopathologic and clinical staging are not sufficient to accurately predict patient prognosis\u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Therefore, it is of utmost urgency to identify new targets for prognostic prediction and treatment of HNSCC.\u003c/p\u003e \u003cp\u003eAbnormal energy metabolism is a fundamental feature of tumors, in which branched-chain amino acid (BCAA, including leucine, valine, isoleucine) metabolism plays a significance role in tumor progression\u003csup\u003e\u003cspan additionalcitationids=\"CR5\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. Cancer cells reshape their metabolism to evade immune surveillance in the tumor microenvironment (TME), forming nutrient-depleted TME and dysfunctional T cells\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Depletion of BCAAs in TME may affect the ability of tumor infiltrating lymphocytes (TILs) to eliminate cancer cells. In the past, the absence of anticipated outcomes from tumor immunotherapy has been intimately linked to the suppressive impact of metabolic reprogramming on immune cells in TME\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. Lymphocyte regulation is influenced by amino acid metabolism. A strong correlation between BCAA intake and total lymphocyte count (TLC) was found in patients with HNSCC in the underweight and elderly groups of the study\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. Several in vitro studies have shown that BCAA deficiency leads to impaired lymphocyte proliferation\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. BCAA metabolism not only enhances energy provision but also actively promotes immune escape of tumor cells\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Leucine is an indispensable substance for the activation of T cells, as a deficiency in leucine inhibits the clonal expansion of Th1, Th17, and CD8\u0026thinsp;+\u0026thinsp;T cell\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Leucine has been discovered to exert an influence on the mammalian target of rapamycin (mTOR) signaling pathway. This pathway is not merely pivotal in orchestrating T-cell differentiation and function; it also wields control over protein translation, as well as cell growth and proliferation\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. BCAA metabolism is associated with the development of many cancers, including breast cancers\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e, glioblastoma\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e, colorectal cancer\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e, and renal cancer\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. It has also been shown that BCAA-MRGs play an indispensable part in prognosticating the outcome of a wide variety of tumors, including multiple myeloma\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e, renal clear cell carcinoma\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e, and pancreatic cancer\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. In summary, BCAA-MRGs show prognostic value in a variety of cancers, demonstrating their great potential for personalized therapy.\u003c/p\u003e \u003cp\u003eTo date, research on BCAA metabolism in HNSCC has been limited. This study explores the potential of BCAA-MRGs as prognostic genes for HNSCC and their possible molecular mechanisms through bioinformatics analysis. Furthermore, single-cell RNA sequencing (scRNA-seq) is used to assess the expression of prognostic genes at the cellular level, and in vitro experiments are conducted to validate the functions of BCAA-MRGs.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eData extraction\u003c/h2\u003e \u003cp\u003eThe Cancer Genome Atlas (TCGA) database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://gdc.cancer.gov/\u003c/span\u003e\u003cspan address=\"https://gdc.cancer.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was applied to obtain the gene expression matrix and clinical data of TCGA-HNSCC dataset, which comprised 502 HNSCC tissue samples and 44 paracancerous tissue samples. Of these, 501 HNSCC samples had complete survival information. This dataset was used as a training set. On the other hand, GSE65858 and GSE140042 datasets were accessed through Gene Expression Omnibus (GEO) database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.ncbi.nlm.nih.gov/geo/\u003c/span\u003e\u003cspan address=\"http://www.ncbi.nlm.nih.gov/geo/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The GSE65858 (platform: GPL10558) dataset was utilized as a validation set, encompassing gene expression matrices and clinical data derived from tumor tissues of 270 HNSCC patients. The scRNA-seq dataset GSE140042 (platform: GPL16791) included tumor tissue samples obtained from 10 HNSCC patients. Additionally, a total of 27 BCAA-MRGs were selected from Molecular Signatures Database (MSigDB) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.gsea-msigdb.org/\u003c/span\u003e\u003cspan address=\"https://www.gsea-msigdb.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) with \u0026lsquo;branched chain amino acids metabolism\u0026rsquo; as a keyword.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eDifferential expression analysis\u003c/h3\u003e\n\u003cp\u003eDifferential expression analysis was carried out using the DEseq2 package\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e (v 3.4.1) to pinpoint differentially expressed genes (DEGs) between tumor and normal tissues within the TCGA-HNSCC dataset. The \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05 \u0026amp; |log2FoldChange (FC)| \u0026gt; 0.5 were utilized as screening criteria. With the application of ggplot2\u003csup\u003e24\u003c/sup\u003e (v 3.4.1) and ComplexHeatmap packages\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e (v 2.14.0), volcano plot and heat map were generated to reveal the expression of DEGs, respectively. Specifically, the heat map exhibited the top10 up-regulated and down-regulated DEGs.\u003c/p\u003e\n\u003ch3\u003eWeighted gene co-expression network analysis (WGCNA)\u003c/h3\u003e\n\u003cp\u003eThe BCAA-MRGs scores for 501 HNSCC samples with complete survival information in the TCGA-HNSCC dataset were computed employing the single-sample gene set enrichment analysis (ssGSEA) algorithm in the GSVA package\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e (v 1.46.0). Meanwhile, 501 HNSCC samples with complete survival information were divided into high and low score groups according to the best cutoff value for BCAA-MRGs scores, and then Kaplan-Meier (K-M) survival analyses were performed using the survminer software package\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e (v 0.4.9) (log-rank test, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) in order to investigate whether there was a difference in survival between the two score groups. WGCNA was conducted using WGCNA package\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e (v 1.70.3) in the 501 HNSCC samples with complete survival information of TCGA-HNSCC dataset, aiming to gain the gene modules associated with BCAA-MRGs scores. To begin with, the HNSCC sample was subjected to a clustering analysis with the objective of assessing the presence of outliers in the sample that needed to be eliminated. Subsequently, the optimal soft threshold power (β) was determined by a scale-free topological fit index greater than 0.85 and a mean connectivity tending towards 0. Following this, the systematic clustering tree was constructed based on the adjacency connection and gene similarity, followed by the construction of a co-expression network using the hybrid dynamic tree cutting algorithm, with a minimum requirement of 100 genes in each gene module. Afterwards, using BCAA-MRGs scores as phenotypic traits, Pearson correlation analysis was performed to calculate the correlation coefficient between modules and BCAA-MRGs scores. The module with the highest correlation with BCAA-MRGs scores was defined as the key module. Furthermore, the genes with |module membership (MM)| \u0026gt; 0.2 and |gene significance (GS)| \u0026gt; 0.2 in the key module were identified as key module genes highly related to BCAA-MRGs scores.\u003c/p\u003e\n\u003ch3\u003eRecognition and functional analysis of candidate genes\u003c/h3\u003e\n\u003cp\u003eCandidate genes were acquired by intersecting DEGs and key module genes using VennDiagram package\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e (v 1.7.3). Based on the GO and KEGG databases, the clusterProfiler package\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e (v 4.7.1.3) was utilized for the enrichment analysis of candidate genes with the aim of exploring the biological functions and signaling pathways of candidate genes in the pathogenesis of HNSCC (\u003cem\u003eP\u003c/em\u003e.adjust\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Furthermore, to gain insight into the interactions between candidate genes, these genes were introduced into the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://string-db.org\u003c/span\u003e\u003cspan address=\"http://string-db.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) website to construct a protein-protein interaction (PPI) network (interaction score\u0026thinsp;\u0026gt;\u0026thinsp;0.4), which was then visualized using Cytoscape software\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e (v 3.7.2) for visualization.\u003c/p\u003e\n\u003ch3\u003eScreening for prognostic genes\u003c/h3\u003e\n\u003cp\u003eIn the 501 HNSCC samples with complete survival information within TCGA-HNSCC dataset, univariate Cox regression analysis was implemented via the survival package\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e (v 3.3-1), resulting in genes significantly associated with the survival of HNSCC [hazard ratio (HR)\u0026thinsp;\u0026ne;\u0026thinsp;1 \u0026amp; \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05]. Subsequently, the forest map was generated for presenting this result using forestplot package\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e (v 2.0.1). Following this, the proportional hazards (PH) hypothesis test was implemented to target these genes that were significantly associated with HNSCC survival (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05). Subsequently, the genes that passed the PH hypothesis test were incorporated into the least absolute shrinkage and selection operator (LASSO) regression analysis. This analysis was executed using the glmnet package\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e (v 4.1.4) to conduct further screening, with the aim of identifying prognostic genes associated with HNSCC. Besides, the expression levels of prognostic genes between tumor and normal groups in TCGA-HNSCC dataset were assessed by the application of the Wilcoxon test (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eConstruction and verification of risk model\u003c/h2\u003e \u003cp\u003e The risk model was constructed based on the identified prognostic genes in the 501 HNSCC samples with complete survival information within TCGA-HNSCC dataset. The risk score was calculated as follows:\u003c/p\u003e \u003cp\u003eriskscore =\u003c/p\u003e \u003cp\u003eIn this formula, βi denoted LASSO regression coefficient, while Expri represented the expression of prognostic gene i. The HNSCC samples were divided into high risk group (HRG) and low risk group (LRG) based on the best cut-off value of risk score. Subsequently, risk curves and survival status maps were plotted, aiming to demonstrate the distribution of patients with HNSCC. Immediately thereafter, survival differences between HRG and LRG were further compared by K-M survival analysis with survminer package (log-rank test, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Furthermore, receiver operating characteristic (ROC) curves were constructed using the survival ROC package\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e (v 1.0.3). Subsequently, area under the curve (AUC) values were calculated. This was done to elucidate the predictive effect of the risk model on the prognosis of HNSCC patients. Importantly, the predictive performance of the risk model was verified in the GSE65858 dataset with the application of risk curves and survival status maps, K-M survival analysis, and ROC curves.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eIndependent prognostic analysis and nomogram evaluation\u003c/h3\u003e\n\u003cp\u003eUnivariate and multivariate Cox regression analyses (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) were implemented for risk score and clinical features (stage, age, gender, T and N stage) to select the independent prognostic factors in the TCGA-HNSCC dataset. After that, a nomogram was constructed employing rms package\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e (v 6.5.0) based on the identified independent prognostic factors with the aim of predicting 1/3/5-year survival in patients with HNSCC. Furthermore, the calibration curves and ROC curves were constructed to assess the predictive capabilities of the nomogram model.\u003c/p\u003e\n\u003ch3\u003eClinical characteristics analysis\u003c/h3\u003e\n\u003cp\u003eStudies have divided all tumors in the TCGA dataset into 6 immune subtypes, namely wound healing (C1), IFN-γ dominant (C2), inflammatory (C3), lymphocyte depleted (C4), immunologically quiet (C5), and TGF-β dominant (C6)\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e,\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. To comprehend the immune subtypes and stage between HRG and LRG, the RColorBrewer package\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e (v 1.1.3) was employed for analyzing the significance of immune subtypes and stage in the TCGA-HNSCC dataset. Additionally, ROC curve analysis was conducted to assess the prognostic predictive ability of risk scores and clinical features for HNSCC patients.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eFunctional enrichment analysis\u003c/h2\u003e \u003cp\u003eIn order to elucidate the biological functional disparities between HRG and LRG of HNSCC patients in TCGA-HNSCC dataset, differential expression analysis was conducted using the DEseq2 package, with c2.kegg.symbols.gmt from MSigDB serving as a reference gene set. Subsequently, genes were ranked based on their calculated log2FC values. Following this step, gene set enrichment analysis (GSEA) was performed utilizing the clusterProfiler package. The screening criteria were defined as |normalized enrichment scores (NES)| \u0026gt; 1, false discovery rate (FDR)\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05. Furthermore, gene set variation analysis (GSVA) was implemented via GSVA package to mine the pathway differences between HRG and LRG, with c2.kegg.symbols.gmt as the reference gene set. Additionally, univariate Cox regression analysis was performed using the survival package to explore the relationship between top 5 pathways obtained from GSVA and survival in HNSCC patients (HR\u0026thinsp;\u0026ne;\u0026thinsp;1 \u0026amp; \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eImmune microenvironment analysis\u003c/h2\u003e \u003cp\u003eTo estimate the TME in patients, we initially computed the stromal score, immune score, and ESTIMATE score for each sample from the TCGA-HNSCC dataset using the ESTIMATE algorithm. Subsequently, we employed the Wilcoxon test to compare any discrepancies in these three scores between HRG and LRG (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). The infiltration scores of the 28 immune cells in each sample were calculated using the ssGSEA algorithm from the GSVA package, and a comparison was made between HRG and LRG by conducting a Wilcoxon test to assess differences in infiltration scores among these immune cells (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Afterwards, Spearman correlation analysis was concluded to reveal the relationship between top five differential immune cells identified according to the significance of the infiltration level and risk score, as well as between differential immune cells and prognostic genes using psych package\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e (v 2.2.9) [|correlation coefficient (cor)|\u0026gt;0.3 \u0026amp; \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05]. Furthermore, HNSCC patients were categorized into high- and low-expression groups based on the optimal cut-off values for differential immune cell expression levels, followed by implementing K-M survival analysis to evaluate survival differences between the high- and low-expression groups (log-rank test, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e \u003cp\u003eOn the other hand, the expression of 24 immune checkpoints was also evaluated between HRG and LRG through Wilcoxon test\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Additionally, the TIDE score, dysfunction score, and exclusion score for HNSCC samples between HRG and LRG were compared by Wilcoxon test (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eSomatic mutation analysis\u003c/h2\u003e \u003cp\u003eIn the TCGA-HNSCC dataset, somatic mutations of patients in HRG and LRG were analyzed using the maftools package\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e (v 2.14.0). In order to evaluate the association between tumor mutational burden (TMB) and patient survival, HNSCC patients within the TCGA-HNSCC dataset were classified into high-TMB and low-TMB groups. This categorization was based on the median value of TMB. Subsequently, K-M survival analysis was employed to explore the survival differences in HNSCC patients among the HRG\u0026thinsp;+\u0026thinsp;high TMB group, HRG\u0026thinsp;+\u0026thinsp;low TMB group, LRG\u0026thinsp;+\u0026thinsp;high TMB group, and LRG\u0026thinsp;+\u0026thinsp;low TMB group.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eDrug sensitivity analysis\u003c/h2\u003e \u003cp\u003eTo identify potential therapeutic drugs for patients in HRG, the oncoPredict package\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e (v 0.2) was employed to predict potential therapeutic drugs in the TCGA-HNSCC dataset. Subsequently, Spearman correlation analysis was performed to assess the association between risk score and the AUC values of these drugs (|cor|\u0026gt;0.3 \u0026amp; \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003escRNA-seq data analysis\u003c/h2\u003e \u003cp\u003eThe expression levels of prognostic genes at the cellular level were assessed in GSE140042 dataset. First, the scRNA-seq data were converted to Seurat objects using the Seurat package\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e (v 5.0.1). The scRNA-seq data were then filtered using the \u0026lsquo;CreateSeuratObject\u0026rsquo; function of the Seurat package, retaining cells with the number of detected genes above 200, followed by calculating the mitochondrial genes using the \u0026lsquo;PercentageFeatureSet\u0026rsquo; function, and retaining cells with a mitochondrial gene percentage of less than 5%. After excluding cells and genes that did not meet the criteria, the data were normalized using the \u0026lsquo;NormalizeData\u0026rsquo; function from the Seurat package with the following parameters: logNormalize, scale.factor\u0026thinsp;=\u0026thinsp;10000. Then, the \u0026lsquo;vst\u0026rsquo; method from \u0026lsquo;FindVariableFeatures\u0026rsquo; function was employed to extract the genes with relatively high coefficients of variation between cells, resulting in the need to find the top 2,000 highly variable genes in this analysis. Following this, principal component analysis (PCA) was examined for the presence of outliers or abnormal cells, followed by the \u0026lsquo;JackStrawPlot\u0026rsquo; function to identify statistically significant principal components (PC) (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). The \u0026lsquo;NormalizeData\u0026rsquo; function was utilized to normalize the data for further clustering of cell clusters. Afterwards, unsupervised cluster analysis was completed using \u0026lsquo;FindNeighbors\u0026rsquo; and \u0026lsquo;FindClusters\u0026rsquo; functions to yield cell clusters (resolution\u0026thinsp;=\u0026thinsp;1). Based on the cell clusters obtained from the above clustering analysis, the marker genes of each cell type were identified by the \u0026lsquo;FindAllMarkers\u0026rsquo; function with the parameters set to only.pos\u0026thinsp;=\u0026thinsp;TRUE and logfc.threshold\u0026thinsp;=\u0026thinsp;1. In conjunction with the auxiliary annotation provided by the \u0026lsquo;SingleR\u0026rsquo; function, the cell clusters were annotated using the related marker genes obtained in the published literature\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e to identify different cell types.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003ePseudo-time analysis for key cells\u003c/h2\u003e \u003cp\u003eThe expression levels of prognostic genes were further compared in different cell types, and then key cells with higher prognostic gene expression were identified. Subsequently, the key cells were subjected to dimensionality reduction and clustering analysis to identify different subpopulations of key cells using the same analysis method as the scRNA-seq data analysis described above. The FindAllMarkers setting parameters min.pct\u0026thinsp;=\u0026thinsp;0.5, logfc.threshold\u0026thinsp;=\u0026thinsp;0.25, test.use\u0026thinsp;=\u0026thinsp;auc were employed to find the marker genes for each subpopulation. Immediately thereafter, the marker genes for each cell subpopulation were compared with the marker genes for each cell type in the CellMarker database to determine the subpopulation types of key cells.\u003c/p\u003e \u003cp\u003eTo explore the state changes of key cells and the expression changes of significantly changed genes, pseudo-time analysis of key subpopulation cells was first performed using Monocle2 package\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e (v 2.22.0). Then, the pseudo-time trajectory was inferred using the \u0026lsquo;reduceDimension\u0026rsquo; function with its default parameters, and the reconstructed trajectory was visualized by utilizing the \u0026lsquo;plot cell trajectory\u0026rsquo; function. Afterwards, 2,000 highly variable genes were extracted, and the genes were clustered according to the pseudo-time expression pattern to generate a representative heat map. Furthermore, the expression of prognostic genes in different developmental states was further analyzed.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eReverse transcription quantitative polymerase chain reaction (RT-qPCR)\u003c/h2\u003e \u003cp\u003eIn an attempt to validate the expression of prognostic genes, 10 HNSCC cancer samples and 10 paracancerous tissue samples were collected from People\u0026rsquo;s Hospital of Xinjiang Uygur Autonomous Region. First, RNA was extracted using 1 ml TRIzol reagent with 50 mg tissue homogenate from each sample, and its concentration was quantified using a NanoPhotometer N50. Then, cDNA was synthesized by reverse transcription according to the SureScript-First-strand-cDNA-synthesis-kit instructions provided by Xavier's. The cDNA was then amplified for 40 cycles using the CFX96 real-time fluorescence quantitative PCR instrument. For RT-qPCR analysis, there were 3 \u0026micro;L cDNA, 5 \u0026micro;L 2x Universal Blue SYBR Green qPCR Master Mix (Servicebio), and 1 \u0026micro;L specific primers per reaction mix. Primer sequences were shown in \u003cb\u003eSupplementary Table\u0026nbsp;1\u003c/b\u003e. Finally, expression of prognostic genes was assessed using the 2-ΔΔCt method.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eStatistical analyses in this study were conducted utilizing R software (v 4.2.3). Differences between the two groups were compared through the implementation of Wilcoxon test. The \u003cem\u003eP\u003c/em\u003e value less than 0.05 was considered statistically significant.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003eRecognition of 8,533 DEGs and 1,369 key module genes\u003c/h2\u003e \u003cp\u003eA total of 8,533 DEGs (tumor vs. normal) were selected through differential expression analysis. Among these, 4,360 genes were found to be up-regulated, while 4,173 genes were down-regulated specifically in the tumor group (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA-B). The Wilcoxon test results revealed differences in BCAA-MRG scores between the tumor and normal groups, with significantly higher scores observed in the normal group. Furthermore, K-M survival analysis between the high and low score groups divided according to the optimal cut-off value of this score demonstrated a significantly worse prognosis for the low score group compared to the high score group (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.0084) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). These findings revealed that BCAA-MRG scores were able to serve as phenotypic traits to implement WGCNA.\u003c/p\u003e \u003cp\u003eThe clustering analysis results indicated that there were no outliers among the 501 HNSCC samples, so all of these samples were included in the subsequent analyses (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD). Then, the optimal soft threshold power (β) was determined to be four based on a scale-free topological fit index greater than 0.85 and a mean connectivity tending towards 0 (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eE). Following this, 11 gene modules were obtained according to the criteria for the construction of a co-expression network (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eF). Among these gene modules, MEyellow (cor=-0.7288, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.98\u0026times;10\u003csup\u003e\u0026minus;\u0026thinsp;20\u003c/sup\u003e) presented the most significant correlation with BCAA-MRG scores and was selected as the key module (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eG). Importantly, 1,369 key module genes were gained based on the screening criteria of |MM| \u0026gt; 0.2 and |GS| \u0026gt; 0.2 (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eH).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eMultiple functions of 257 candidate genes with interactions were characterized\u003c/h2\u003e \u003cp\u003eA total of 257 candidate genes were identified by crossing 8,533 DEGs and 1,369 key module genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). The signaling biological functions and pathways in which these candidate genes were significantly enriched were further explored by enrichment analyses. The results revealed that these candidate genes were significantly enriched to 146 GO entries, which contained 26 molecular functions, 70 cellular components, and 50 biological processes, as well as 31 KEGG pathways. According to the \u003cem\u003eP\u003c/em\u003e.adjust value ranking, the enriched top10 GO entries (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB) and top10 KEGG pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC) were selected for presentation, such as aerobic respiration, ATP synthesis coupled electron transport, Chemical carcinogenesis-reactive oxygen species, oxidative phosphorylation, respiratory electron transport chain, thermogenesis, etc. These results implicated that the candidate genes might be influencing the development of HNSCC by affecting the energy metabolism and conversion of cells. Additionally, the results of the PPI network demonstrated that UQCRFS1, UQCRC2, UQCRC1, COX7C, NDUFA4, and SNRPE interacted with more genes (\u003cb\u003eSupplementary Fig.\u0026nbsp;1\u003c/b\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eSMS, PRDX6, GSTO1, and ADA were confirmed as prognostic genes to construct the risk model\u003c/h2\u003e \u003cp\u003eSix genes significantly associated with HNSCC survival were identified, namely PFDN4, SMS, PRDX6, GSTO1, COA6, and ADA, employing univariate Cox regression analysis, and all of these genes were risk factors for HNSCC (HR\u0026thinsp;\u0026gt;\u0026thinsp;1) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). Meanwhile, these genes all passed the PH hypothesis test (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05) (\u003cb\u003eSupplementary Fig.\u0026nbsp;2A\u003c/b\u003e). In the LASSO regression analysis, SMS, PRDX6, GSTO1, and ADA were identified and recorded as prognostic genes (lambda.min\u0026thinsp;=\u0026thinsp;0.01019419) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB). Furthermore, the expression assessment of these prognostic genes indicated that PRDX6 was downregulated in the tumor group compared to the normal group, whereas SMS, GSTO1, and ADA were upregulated (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). Similarly, in the RT-qPCR experiment, SMS, GSTO1, and ADA all expressed highly in the tumor group (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD-F). However, there was no remarkably difference in PRDX6 between the tumor and the normal groups (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026gt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eG). Based on the expression levels of the four prognostic genes and the corresponding coefficients in LASSO regression analysis, a risk model was constructed utilizing the following formula: risk score\u0026thinsp;=\u0026thinsp;0.2138\u0026times;SMS expression level\u0026thinsp;+\u0026thinsp;0.2144 \u0026times; PRDX6 expression level\u0026thinsp;+\u0026thinsp;0.0512 \u0026times; GSTO1 expression level\u0026thinsp;+\u0026thinsp;0.1510 \u0026times; ADA expression level.\u003c/p\u003e \u003cp\u003eThe HNSCC samples with complete survival information were classified into HRG and LRG based on the best cut-off value of risk score. As demonstrated in \u003cb\u003eSupplementary Fig.\u0026nbsp;2B\u003c/b\u003e, it could be found that the mortality rate of patients increased with the increase of risk score. K-M survival analysis confirmed significant differences in survival between HRG and LRG, with HRG having significantly worse survival compared to LRG (\u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.0001) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eH). ROC curve indicated an appropriate accuracy of risk model in predicting survival, with AUC values of 0.615, 0.631, 0.614 at 1-, 3-, and 5-year, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eI). Furthermore, the robust prediction ability of risk model was validated in GSE65858 dataset, yielding consistent results. Specifically, the risk curves and survival status maps revealed that as the risk score increased, the mortality rate of patients also increased. K-M survival analysis confirmed that LRG patients had notably lower survival than HRG patients. Also, the AUC values in the ROC curves were all greater than 0.6 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eJ-K, \u003cb\u003eSupplementary Fig.\u0026nbsp;2C\u003c/b\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e \u003ch2\u003eExploration of predictive performance for nomogram and clinical features\u003c/h2\u003e \u003cp\u003eTwo independent prognostic factors, risk score and age, were generated through the application of univariate and multivariate Cox logistic regression analyses (Fig.\u0026nbsp;4A-B). The nomogram model was constructed using risk score and age to predict 1/3/5-year survival in patients with HNSCC (Fig.\u0026nbsp;4C). The calibration curves indicated that the slopes of the nomogram model at 1/3/5 years were close to the theoretical line, indicating a high level of predictive accuracy for the nomogram model (Fig.\u0026nbsp;4D). Meanwhile, the AUC values of the nomogram model were consistently above 0.6 at 1 (0.894), 3 (0.678), and 5 (0.678) years, revealing the excellent predictive ability of the nomogram model for the survival of HNSCC patients (Fig.\u0026nbsp;4E). On the other hand, the differences of immune subtypes and stage between HRG and LRG were compared. The results revealed that the immune subtypes were not significantly different between HRG and LRG (P\u0026thinsp;=\u0026thinsp;0.088) (\u003cb\u003eSupplementary Fig.\u0026nbsp;3A\u003c/b\u003e), whereas stage was significantly different (P\u0026thinsp;=\u0026thinsp;0.013) between HRG and LRG (\u003cb\u003eSupplementary Fig.\u0026nbsp;3B\u003c/b\u003e). Furthermore, the ROC curve analysis demonstrated that the AUC values at 1/3/5 years were greater than 0.6 for both risk score and N stage, suggesting a better prognostic ability for risk score and N stage \u003cb\u003e(Supplementary Fig.\u0026nbsp;3C).\u003c/b\u003e\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003eSignaling pathways significantly associated with HNSCC prognosis were identified\u003c/h2\u003e \u003cp\u003eGSEA results indicated that top 5 signaling pathways significantly enriched for DEGs between HRG and LRG were \u0026lsquo;ribosome\u0026rsquo;, \u0026lsquo;huntingtons disease\u0026rsquo;, \u0026lsquo;dilated cardiomyopathy\u0026rsquo;, \u0026lsquo;hypertrophic cardiomyopathy hcm\u0026rsquo;, and \u0026lsquo;focal adhesion\u0026rsquo; (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). At the same time, GSVA results showed that \u0026lsquo;notch signaling pathway\u0026rsquo;, \u0026lsquo;fructose and mannose metabolism\u0026rsquo;, \u0026lsquo;pentose phosphate pathway\u0026rsquo;, \u0026lsquo;natural killer cell-mediated cytotoxicity\u0026rsquo;, \u0026lsquo;regulation of autophagy\u0026rsquo; and other pathways were significantly different between HRG and LRG (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). Moreover, univariate Cox regression analysis was conducted to explore the relationship between the top five pathways obtained from GSVA and survival in HNSCC patients. The results demonstrated that these five pathways were significantly related to the survival of HNSCC patients. \u003c/p\u003e \u003cp\u003e \u003cb\u003eFigure 4\u003c/b\u003e The clinical utility of risk modeling. Univariate (A) and multivariate (B) Cox analysis based on risk scores and other clinical characteristics. (C) The nomogram for predicting the survival probability of HNSCC patients has two independent prognostic features. (D) The calibration plots of the nomogram for predicting OS probability for 1-, 3-, and 5-years. (E) ROC analysis of the nomogram.\u003c/p\u003e \u003cp\u003eWith the exception of the JAK/STAT signaling pathway, all other signaling pathways facilitated HNSCC progression (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003eC).\u003c/p\u003e \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e \u003ch2\u003eRisk score was strongly associated with the immune microenvironment in HNSCC patients\u003c/h2\u003e \u003cp\u003eESTIMATE algorithm results revealed considerable disparities in stromal scores, immune scores, and ESTIMATE scores between HRG and LRG, with significantly higher values observed in the LRG (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003eA). Meanwhile, the infiltration scores of 20 out of 28 immune cells were found to be considerably different between HRG and LRG. All immune cells except CD56bright natural killer cell had higher infiltration scores in LRG, including monocytes, neutrophils, and eosinophils (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003eB). The\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003etop five differential immune cells were plasmacytoid dendritic cell, T follicular helper cell, CD56bright natural killer cell, type 17 T helper cell, and mast cell (\u003cb\u003eSupplementary Table\u0026nbsp;2A)\u003c/b\u003e. Correlation analysis confirmed that CD56bright natural killer cell showed a significant positive correlation with the risk score, while other four immune cells were opposite (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003eC). Furthermore, significant differences in the survival of T follicular helper cell, Type 17 T helper cell, CD56bright natural killer cell, and mast cell between HRG and LRG revealed that all of these immune cells were associated with the survival of HNSCC (\u003cb\u003eSupplementary Fig.\u0026nbsp;4A-E\u003c/b\u003e). Additionally, a noteworthy inverse correlation was observed between PRDX6 and plasmacytoid dendritic cell cells (cor=-0.36, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;3.95\u0026times;10\u003csup\u003e\u0026minus;\u0026thinsp;17\u003c/sup\u003e), while a significant positive correlation was found between GSTO1 and CD56bright natural killer cells (cor\u0026thinsp;=\u0026thinsp;0.37, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;7.9\u0026times;10\u003csup\u003e\u0026minus;\u0026thinsp;17\u003c/sup\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003eD, \u003cb\u003eSupplementary Table\u0026nbsp;2B\u003c/b\u003e).\u003c/p\u003e \u003cp\u003eOn the other hand, a significant disparity was observed in the expression levels of 10 immune checkpoints between the HRG and the LRG, namely CD276, CD27, CD40, CD70, ENTPD1, ICOS, SIGLEC15, TNFRSF18, TNFRSF9, TNFSF9. Among them, CD276, CD40, CD70, TNFRSF18, and TNFSF9 were significantly higher expressed in HRG, while the opposite was true for the remaining five immune checkpoints (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003eE). Also, the TIDE score, dysfunction score, and exclusion score for HNSCC samples between HRG and LRG were compared. The results demonstrated that there were no statistically considerable disparities in TIDE score between HRG and LRG. However, LRG exhibited significantly higher sensitivity to dysfunction scores, while HRG showed significantly greater sensitivity to exclusion scores (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003eF).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e \u003ch2\u003eThe identification of drugs linked to risk score and the assessment of TMB\u003c/h2\u003e \u003cp\u003eDrug sensitivity analysis revealed that there was a considerable association between 15 drugs and risk scores, namely doramapimod_1042, GDC0810_1925, KU-55933_1030, NU7441_1038, pevonedistat_1529, GSK591_2110, rapamycin_1084, ZM447439_1050, AZD6738_1917, paclitaxel_1080, AZD7762_1022, docetaxel_1007, ABT737_1910, JQ1_2172, and AZD8186_1918. Importantly, doramapimod_1042 was considerably positively associated with the risk score (cor\u0026thinsp;=\u0026thinsp;0.43, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;2.2\u0026times;10\u003csup\u003e\u0026minus;\u0026thinsp;16\u003c/sup\u003e), while docetaxel_1007 had a considerable negative relationship with risk score (cor=-0.33, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;1.4\u0026times;10\u003csup\u003e\u0026minus;\u0026thinsp;13\u003c/sup\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e7\u003c/span\u003eA).\u003c/p\u003e \u003cp\u003eIn the TCGA-HNSCC dataset, somatic mutations of patients in HRG and LRG were analyzed. The results indicated that TP53 and TTN were the top two mutated genes in both HRG and LRG (\u003cb\u003eSupplementary Fig.\u0026nbsp;5A-B\u003c/b\u003e). In addition, we observed a significant increase in TMB values in HRG (\u003cb\u003eSupplementary Fig.\u0026nbsp;5C\u003c/b\u003e) and found a strong correlation between these values and survival outcomes among patients with HNSCC (\u003cb\u003eSupplementary Fig.\u0026nbsp;5D\u003c/b\u003e). These results indicated that the involvement of TMB was significant in the progression of HNSCC.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section3\"\u003e \u003ch2\u003ePrognostic genes expression changed with different stages of cell differentiation\u003c/h2\u003e \u003cp\u003eIn the GSE140042 dataset, there were 21,489 cells and 33,538 genes prior to QC (\u003cb\u003eSupplementary Fig.\u0026nbsp;6A\u003c/b\u003e), whereas post-QC analysis generated 11,631 cells and 33,538 genes (\u003cb\u003eSupplementary Fig.\u0026nbsp;6B\u003c/b\u003e). Then, the \u0026ldquo;vst\u0026rdquo; method was applied to obtain the top 2,000 highly variable genes (\u003cb\u003eSupplementary Fig.\u0026nbsp;6C\u003c/b\u003e). PCA results\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eshowed no significant outliers in the samples from GSE140042 dataset (\u003cb\u003eSupplementary Fig.\u0026nbsp;6D\u003c/b\u003e). Following this, 20 PCs were identified for subsequent analysis (\u003cb\u003eSupplementary Fig.\u0026nbsp;6E\u003c/b\u003e). Immediately after, 30 cell clusters were obtained by clustering analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eA), followed by demonstrating the expression of marker genes in these cell clusters (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eB). Furthermore, six cell types were annotated, including T cells, CD4\u003csup\u003e+\u003c/sup\u003e T cells, fibroblasts, CD8\u0026thinsp;+\u0026thinsp;T cells, B cells, and epithelial cells (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eC). Importantly, epithelial cells were selected as key cells due to higher expression of prognostic genes (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eD).\u003c/p\u003e \u003cp\u003eNine subtypes were obtained by dimensionality reduction (\u003cb\u003eSupplementary Fig.\u0026nbsp;6F\u003c/b\u003e) and clustering analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eE) of epithelial cells, followed by annotation to determine the specific cell types of these subtypes (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eF). Pseudo-time analysis for key cells revealed that the cell subtypes of epithelial cells, urothelium cell, premeiotic germ cell and endoderm progenitor cell, were at state1, state4 and state5 of epithelial cell differentiation, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eG). Then, 2,000 marker genes with significant changes were extracted, and the genes were clustered to three clusters. The heat map revealed the changes in expression of these three gene clusters (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eH). Moreover, the expression of ADA and PRDX6 demonstrated a temporal pattern. As the pseudo-time changed, their expression first decreased and then increased. Conversely, GSTO1 displayed an opposing trend; its expression initially increased and subsequently decreased. In stark contrast, SMS showed no substantial change in its expression over the pseudo-time scale. (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e8\u003c/span\u003eI).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eHNSCC is a type of common malignant tumor, which seriously threatens the life and health of patients. As essential amino acids, BCAAs have attracted much attention in recent years in the field of tumor research, including their effects on tumor cell metabolism, signaling pathways, immune regulation, and TME\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e,\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e.The study combined RNA-seq and scRNA-seq to construct and validate an BCAA-MRGs risk model. The results showed that the model could accurately predict the prognosis of HNSCC patients, and the expression levels of prognostic genes were verified by RT-qPCR. These findings provide directions for the specific treatment of HNSCC patients.\u003c/p\u003e \u003cp\u003ePrevious studies have shown that BCAA metabolism affects different cellular processes, ranging from protein synthesis to epigenetics\u003csup\u003e\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e. Disturbed BCAA metabolism in cancer may affect these processes and lead to disease progression, potentially serving as a marker for predicting disease status. Therefore, it is crucial to study the genes associated with BCAA metabolism in HNSCC. In this study, SMS, PRDX6, GSTO1 and ADA were identified as prognostic genes for risk modeling. The HNSCC samples were categorized into HRG and LRG based on the optimal threshold of the risk score, with LRG having a significantly higher overall survival rate than HRG. The SMS (spermine synthase) gene, a member of the spermidine/spermine synthase family, is located on chromosomes 1, 5, 6, and X\u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e. Recent studies have shown that SMS is associated with the progression of a variety of cancers, including hepatocellular carcinoma\u003csup\u003e\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e\u003c/sup\u003e, breast cancer\u003csup\u003e\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u003c/sup\u003e, colorectal cancer\u003csup\u003e\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e, and HNSCC\u003csup\u003e\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e. Pan et al. found that SMS expression was elevated in HNSCC compared to normal tissues, and poor overall survival was associated with high SMS expression\u003csup\u003e\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e. These findings are consistent with ours. Upregulation of SMS expression leads to increased intracellular levels of spermine, one of the polyamines, which promotes tumor cell invasion, proliferation and metastasis while inhibiting apoptosis. In tumor cells, the mTOR signaling pathway is both affected by polyamine levels and involved in the regulation of BCAA metabolism\u003csup\u003e\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e,\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e\u003c/sup\u003e. Polyamine may indirectly affect the expression of BCAA-MRGs and enzyme activities by regulating the activity of the mTOR signaling pathway, thus forming a complex metabolic regulatory network in tumor cells.\u003c/p\u003e \u003cp\u003ePRDX6, a member of the antioxidant enzyme superfamily, plays an important role in cell differentiation, apoptosis, and redox-mediated signaling\u003csup\u003e\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e\u003c/sup\u003e. Data analysis showed that PRDX6 expression was down-regulated in HNSCC tumor tissues compared with normal tissues, and the expression trend of PRDX6 was found to be consistent in HNSCC tumor tissues by RT-qPCR, although there was no significant difference in the expression, which might be related to the heterogeneity of the samples. PRDX6 was found to be a poor prognostic factor for tongue squamous cell carcinoma, and its high expression was significantly correlated with shortened OS in patients\u003csup\u003e\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e\u003c/sup\u003e. In addition, PRDX6 expression was up-regulated in the subgroup with poor histopathological differentiation of HNSCC, and the high expression group showed reduced OS\u003csup\u003e\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e\u003c/sup\u003e, suggesting that this gene is an independent prognostic factor for HNSCC. This is consistent with our findings, where our univariate Cox regression analysis results showed that PRDX6 is a risk factor for HNSCC. It has been shown that in HNSCC, PRDX6 inhibits apoptosis of tumor cells by exerting oxidative effects\u003csup\u003e\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e\u003c/sup\u003e. Given the important role of PRDX6 in HNSCC, it is expected to be a potential therapeutic target.\u003c/p\u003e \u003cp\u003eGSTO1 (Glutathione S-transferase omega 1) is a multifunctional enzyme with detoxification, redox regulation, and is involved in the regulation of signaling pathways in a variety of pathological diseases including cancer. GSTO1 expression is up-regulated in many cancers, such as bladder cancer \u003csup\u003e\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e\u003c/sup\u003e, non-small cell lung cancer \u003csup\u003e\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e\u003c/sup\u003e, and esophageal squamous cell carcinomas \u003csup\u003e\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e\u003c/sup\u003e, and is in connection with poor prognosis. In the present study, we also found that high expression of GSTO1 was associated with poor prognosis in HNSCC. GSTO1 plays an important role in the regulation of intracellular redox state. It was found that, on the one hand, the energy and metabolites generated by BCAA metabolism can regulate the level of reactive oxygen species (ROS) and affect cell proliferation. On the other hand, ROS can alter the supply and metabolism of BCAA by affecting BCAA metabolism-related enzymes and transporter proteins, which in turn affects the activity of the mTOR signaling pathway and promotes or inhibits tumor proliferation\u003csup\u003e\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e,\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e\u003c/sup\u003e. Accordingly, we hypothesize that GSTO1 may affect the metabolism of BCAA through regulating the level of intracellular ROS and influence the development of tumors.\u003c/p\u003e \u003cp\u003eAdenosine deaminase (ADA) contains two isoenzymes, ADA1 and ADA2, which are mainly generated from adenosine triphosphate (ATP) through a series of enzymatic reactions, and adenosine is an important immunosuppressive signal\u003csup\u003e\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e,\u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e66\u003c/span\u003e\u003c/sup\u003e. ADA affects cell proliferation, differentiation, and apoptosis by regulating intracellular adenosine levels. Research has found that elevated BCAA levels alter ATP production, interfering with glycolysis, fatty acid oxidation, the tricarboxylic acid (TCA) cycle, and oxidative phosphorylation, leading to mitochondrial dysfunction\u003csup\u003e\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e\u003c/sup\u003e. Adenosine may be the key step linking ADA to BCAA metabolism. In summary, there is a close and complex interaction between BCAA metabolism and tumors, and this interaction plays an important role in tumor development by affecting BCAA metabolism-related enzymes and transport proteins. It is expected that further in-depth study of BCAA metabolism to provide new targets and strategies for the diagnosis and treatment of patients with HNSCC, and to bring hope for the improvement of the prognosis of patients with HNSCC.\u003c/p\u003e \u003cp\u003eTo further explore the pathogenic mechanism of HNSCC, we performed GSEA and GSVA pathway enrichment analysis and univariate Cox regression analysis of GSVA enrichment results, and found that \u0026ldquo;fructose and mannose metabolism\u0026rdquo;, \u0026ldquo;pentose phosphate pathway\u0026rdquo; and \u0026ldquo;natural killer cell-mediated cytotoxicity\u0026rdquo; as well as Wnt signaling pathway were upregulated and significantly connected with poor prognosis of HNSCC. It is well known that many cancers exhibit a strong glycolytic phenotype, which usually correlates with the aggressiveness and malignancy of the tumor\u003csup\u003e\u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e,\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e\u003c/sup\u003e. In cancer, cytoplasmic glucose can also be utilized through the pentose phosphate pathway, which induces cell proliferation through the synthesis of nucleic acids, fatty acids, and amino acids\u003csup\u003e\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e\u003c/sup\u003e. Jonathan et al. found higher glycolysis scores in HNSCC by RNA-seq profiling of glycolytic tumors in a pan-cancer cohort\u003csup\u003e\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e\u003c/sup\u003e. This is consistent with the results of the present study that glycolysis may induce amino acid synthesis in HNSCC by interacting with the pentose phosphate pathway, thereby promoting cancer cell proliferation. Therefore, the development of tumor therapeutic strategies targeting this metabolic pathway will offer novel strategies for the treatment of HNSCC. In addition, this study found that natural killer cell-mediated cytotoxicity pathway was upregulated and associated with poor prognosis of HNSCC. While the conventional view is that natural killer cells inhibit tumor growth through their cytotoxic effects, recent studies have found that NK cell-mediated cytotoxicity may promote tumor progression under certain circumstance\u003csup\u003e\u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e72\u003c/span\u003e\u003c/sup\u003e. For example, NK cells can not only promote epithelial-mesenchymal transition (EMT) of tumor cells by secreting cytokines and chemokines to enhance tumor migration and invasion, but also inhibit anti-tumor immune responses by attracting immunosuppressive cells into the TME, such as regulatory T cells.\u003c/p\u003e \u003cp\u003eTo further reveal the TME in HNSCC patients, we performed a series of TME correlation analyses. In the TME, there is a competitive relationship between immune cells and tumor cells for BCAAs. Immune cells can absorb and utilize BCAAs, inhibiting the progress of the tumor. When tumor cells consume a large amount of BCAAs, the access of immune cells to BCAAs can be inhibited, resulting in immune evasion\u003csup\u003e\u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e73\u003c/span\u003e\u003c/sup\u003e. In our study, we found significant differences in the TME between HRG and LRG. Among them, plasmacytoid dendritic cells, mast cells, type 17 T helper cells, and T follicular helper cells were infiltrated to a higher extent in the LRG than in the HRG, while CD56 bright natural killer cells were infiltrated to a low extent. Wulff et al. found lower numbers of circulating immunoregulatory CD56 bright natural killer cells in the peripheral blood of patients with HNSCC, regardless of individual tumor stage or tumor type\u003csup\u003e\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e\u003c/sup\u003e. This coincides with the results of the present study. The high - uptake of tumor cells can lead to local changes in the concentration of BCAAs in the TME. BCAAs are important nutrients for NK cell metabolism. In the absence of BCAAs, the metabolism of NK cells will be disrupted. CD56\u0026thinsp;+\u0026thinsp;natural killer (NK) cells are essential for innate antiviral and antitumor host defense. Based on their CD56 surface density expression, they can be categorized into dim and bright subpopulations. Dim NK cells are cytolytic and account for approximately 90% of NK cells, whereas bright NK cells are immunomodulatory\u003csup\u003e\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e\u003c/sup\u003e. Lower infiltration of CD56 bright natural killer cells leads to decreased immunomodulation. In addition, this study found that the prognostic gene GSTO1 was significantly and positively correlated with CD56 bright natural killer cells. GSTO1 has the potential to affect the balance of activating and inhibiting receptor expression on the surface of NK cells, thereby altering the functional status of NK cells and influencing the progression of HNSCC. The remarkable plasticity of Th17 cells, a subset of T cells that can produce IL-22, IL-21 IL-17A, IL-17F, and IFN-γ, is an important player in inflammatory and autoimmune diseases as well as cancer\u003csup\u003e\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e76\u003c/span\u003e,\u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e77\u003c/span\u003e\u003c/sup\u003e. In a study on oral cancer, it was found that mice with oral cancer had a low Th17 phenotype and that the Th17 phenotype declined with cancer progression\u003csup\u003e\u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e78\u003c/span\u003e\u003c/sup\u003e. Mast cells can remodel the TM, which may change the accessibility of tumor cells and other cells to BCAAs by influencing angiogenesis and nutrient transport in the TME. Mast cells can directly exert cytotoxic effects and induce apoptosis of tumor cells, and recruit and activate immune cells, including NK cells and cytotoxic T lymphocytes, to the tumor site, thus exerting anti-tumor effects\u003csup\u003e\u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e79\u003c/span\u003e\u003c/sup\u003e. In summary, immune cells have a complex role in HNSCC, promoting both tumor progression and anti-tumor activity. The present study is expected to enhance the anti-tumor function of immune cells by modulating the BCAA-related metabolic pathways to bring better therapeutic outcomes for HNSCC patients.\u003c/p\u003e \u003cp\u003eIn order to find potential drugs for the treatment of HNSCC patients, this study analyzed the drug sensitivity of a specific subgroup of HRG and found that they were sensitive to 15 drugs, including doramapimod, rapamycin, paclitaxel, and docetaxel. Doramapimod is a small molecule inhibitor of p38 mitogen-activated protein kinase (MAPK). The p38 MAPK signaling pathway plays a key role in tumor-related inflammation and immune regulation. Studies have found increased sensitivity to doramapimod chemotherapy regimens in esophageal squamous cell carcinoma\u003csup\u003e\u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e80\u003c/span\u003e\u003c/sup\u003e. Docetaxel belongs to the paclitaxel class of drugs, and its main mechanism of action is to promote the assembly of microtubule proteins into stable microtubules and inhibit microtubule depolymerization, which stops cell mitosis in the G2-M phase, thus inhibiting the proliferation of tumor cells. Docetaxel-platinum-cetuximab treatment showed promising results in a phase 2 trial of first-line recurrent or metastatic HNSCC\u003csup\u003e\u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e81\u003c/span\u003e\u003c/sup\u003e. Paclitaxel in combination with cisplatin or carboplatin is one of the classic chemotherapeutic regimens in the treatment of HNSCC. Paclitaxel acts on the microtubule system and serves to kill tumor cells\u003csup\u003e\u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e82\u003c/span\u003e\u003c/sup\u003e. In the future, with in-depth research on the pathogenesis of HNSCC and advances in drug development, chemotherapy treatment regimens are expected to be continuously optimized, bringing better therapeutic effects and quality of life to HNSCC patients.\u003c/p\u003e \u003cp\u003eIn order to reveal the detailed heterogeneity of prognostic genes in tumor cells and to compare the expression of prognostic genes in various cell clusters, we performed scRNA-seq analysis, and found that prognostic genes were expressed in CD4\u0026thinsp;+\u0026thinsp;T cells, fibroblasts, and CD8\u0026thinsp;+\u0026thinsp;T cells, especially in epithelial cells. It was found that in HNSCC, epithelial cells could interact with immune cells and fibroblasts to co-construct a microenvironment that promotes tumor growth and supports tumor progression\u003csup\u003e\u003cspan citationid=\"CR83\" class=\"CitationRef\"\u003e83\u003c/span\u003e,\u003cspan citationid=\"CR84\" class=\"CitationRef\"\u003e84\u003c/span\u003e\u003c/sup\u003e. Dysfunctional epithelial cells can enhance the invasiveness of migrating HNSCC tumor cells\u003csup\u003e\u003cspan citationid=\"CR85\" class=\"CitationRef\"\u003e85\u003c/span\u003e,\u003cspan citationid=\"CR86\" class=\"CitationRef\"\u003e86\u003c/span\u003e\u003c/sup\u003e. Different subpopulations of epithelial cells were further analyzed, and it was found that prognostic genes were differently expressed in different developmental states, and the expression levels of two genes, ADA and PRDX6, decreased and then increased with pseudo time, while GSTO1 increased and then decreased, and SMS showed an overall unchanged trend.\u003c/p\u003e \u003cp\u003eIn this study, we explored the role of BCAAs in HNSCC by combining RNA-seq and scRNA-seq data, and identified four key genes related to BCAA metabolism, based on which we constructed, for the first time, a reliable model that can predict the prognosis of HNSCC. In addition, we found that the key genes may be associated with the sensitivity of HNSCC patients to chemotherapeutic agents and immune checkpoint inhibitors. Moreover, we also found that in epithelial cells, the expression levels of key genes vary with different stages of cell differentiation. And the expression levels of the key genes were further validated experimentally. The identification of these key genes provides important clinical implications for the diagnosis and treatment of HNSCC. However, our study is inevitably subject to certain limitations. First, the main data for our study were obtained from TCGA and GEO databases. Second, the number of tumor samples we used to validate prognostic genes was small. In the future, we will continue to further explore the mechanism of branched-chain amino acid metabolism-related prognostic genes in HNSCC in various aspects of clinical and laboratory studies.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn summary, this study developed a reliable prognostic model based on BCAAs. And initially but relatively comprehensively explored the potential mechanisms by which these genes influence patient prognosis. This provides a new reference for the mechanism of action of BCAA-MRGs in HNSCC.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData supporting the results of this study are publicly available from the TCGA databases (TCGA-HNSCC, https://gdc.cancer.gov/) and GEO databases (GSE65858 and GSE140042, https://www.ncbi.nlm. nih.gov/geo/). The experimental data will be made available from the corresponding author upon reasonable request.We thank all the individuals who participated in this study. We also thank the GEO and TCGA data platform. In Addition, we sincerely thank the Department of Otorhinolaryngology and Head and Neck Surgery, Xinjiang Uygur Autonomous Region People's Hospital for providing the study samples of squamous cell carcinoma of the head and neck.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData supporting the results of this study are publicly available from the TCGA databases (TCGA-HNSCC, https://gdc.cancer.gov/) and GEO databases (GSE65858 and GSE140042, https://www.ncbi.nlm. nih.gov/geo/). The experimental data will be made available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization, D.L. and A.A.; methodology, D.L. and Y.L.; resources, D.L. and A.A.; writing\u0026mdash;original draft preparation, D.L. and T.M.; writing\u0026mdash;review and editing, D.L., Y.L. and A.A.; visualization, Y.L. and T.M.; supervision, D.L. and A.A.; project administration, A.A.; funding acquisition, D.L. and A.A. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Natural Science Foundation of Xinjiang Uygur Autonomous Region (2021D01C188) and the Science Foundation of 215 Hospital of Shaanxi Nuclear Industry (215KYJJ-202324).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthical approval statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data for this study were obtained from the open databases TCGA and GEO. Cancer tissues and paracancerous tissues are obtained from patients with Head and neck squamous cell carcinoma in the Xinjiang Uygur Autonomous Region People's Hospital. All experimental protocols were approved by the Research Ethics Committee of Xinjiang Uygur Autonomous Region People's Hospital.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of Interest Statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests in this work.\u003c/p\u003e\n"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBray, F. et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. \u003cem\u003eCA Cancer J. Clin.\u003c/em\u003e \u003cb\u003e74\u003c/b\u003e, 229\u0026ndash;263. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3322/caac.21834\u003c/span\u003e\u003cspan address=\"10.3322/caac.21834\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao, X., Chen, H., Qiu, Y. \u0026amp; Cui, L. FAM64A promotes HNSCC tumorigenesis by mediating transcriptional autoregulation of FOXM1. \u003cem\u003eInt. J. Oral Sci.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41368-022-00174-4\u003c/span\u003e\u003cspan address=\"10.1038/s41368-022-00174-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNishimura, A. et al. Incurable locoregional disease is a strong poor prognostic factor in recurrent or metastatic squamous cell carcinoma of the head and neck. \u003cem\u003eInt. J. Clin. Oncol.\u003c/em\u003e \u003cb\u003e26\u003c/b\u003e, 1822\u0026ndash;1830. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s10147-021-01965-1\u003c/span\u003e\u003cspan address=\"10.1007/s10147-021-01965-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, B. et al. Association of dietary intake of branched-chain amino acids with long-term risks of CVD, cancer and all-cause mortality. \u003cem\u003ePublic Health. Nutr.\u003c/em\u003e \u003cb\u003e25\u003c/b\u003e, 1\u0026ndash;11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1017/s1368980021004948\u003c/span\u003e\u003cspan address=\"10.1017/s1368980021004948\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNouri-Majd, S., Salari-Moghaddam, A., Benisi-Kohansal, S., Azadbakht, L. \u0026amp; Esmaillzadeh, A. Dietary intake of branched-chain amino acids in relation to the risk of breast cancer. \u003cem\u003eBreast cancer (Tokyo Japan)\u003c/em\u003e. \u003cb\u003e29\u003c/b\u003e, 993\u0026ndash;1000. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s12282-022-01379-5\u003c/span\u003e\u003cspan address=\"10.1007/s12282-022-01379-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSciacovelli, M. et al. Dynamic partitioning of branched-chain amino acids-derived nitrogen supports renal cancer progression. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 7830. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41467-022-35036-4\u003c/span\u003e\u003cspan address=\"10.1038/s41467-022-35036-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim, S. K. \u0026amp; Cho, S. W. The Evasion Mechanisms of Cancer Immunity and Drug Intervention in the Tumor Microenvironment. \u003cem\u003eFront. Pharmacol.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 868695. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fphar.2022.868695\u003c/span\u003e\u003cspan address=\"10.3389/fphar.2022.868695\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAktar, N. et al. Understanding of Immune Escape Mechanisms and Advances in Cancer Immunotherapy. \u003cem\u003eJournal of oncology\u003c/em\u003e 8901326, (2022). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1155/2022/8901326\u003c/span\u003e\u003cspan address=\"10.1155/2022/8901326\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDePeaux, K. \u0026amp; Delgoffe, G. M. Metabolic barriers to cancer immunotherapy. \u003cem\u003eNat. Rev. Immunol.\u003c/em\u003e \u003cb\u003e21\u003c/b\u003e, 785\u0026ndash;797. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41577-021-00541-y\u003c/span\u003e\u003cspan address=\"10.1038/s41577-021-00541-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKristian, Y. Y. et al. Correlation between branched-chain amino acids intake and total lymphocyte count in head and neck cancer patients: a cross-sectional study. \u003cem\u003eBMC Nutr.\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e, 86. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s40795-023-00746-5\u003c/span\u003e\u003cspan address=\"10.1186/s40795-023-00746-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDauphinais, C. \u0026amp; Waithe, W. I. PHA stimulation of human lymphocytes during amino acid deprivation. Protein, RNA and DNA synthesis. \u003cem\u003eJ. Cell. Physiol.\u003c/em\u003e \u003cb\u003e91\u003c/b\u003e, 357\u0026ndash;367. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/jcp.1040910305\u003c/span\u003e\u003cspan address=\"10.1002/jcp.1040910305\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (1977).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChuang, J. C., Yu, C. L. \u0026amp; Wang, S. R. Modulation of human lymphocyte proliferation by amino acids. \u003cem\u003eClin. Exp. Immunol.\u003c/em\u003e \u003cb\u003e81\u003c/b\u003e, 173\u0026ndash;176. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/j.1365-2249.1990.tb05310.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1365-2249.1990.tb05310.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (1990).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWetzel, T. J., Erfan, S. C. \u0026amp; Ananieva, E. A. The emerging role of the branched chain aminotransferases, BCATc and BCATm, for anti-tumor T-cell immunity. \u003cem\u003eImmunometabolism (Cobham Surrey)\u003c/em\u003e. \u003cb\u003e5\u003c/b\u003e, e00014. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/in9.0000000000000014\u003c/span\u003e\u003cspan address=\"10.1097/in9.0000000000000014\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSinclair, L. V. et al. Control of amino-acid transport by antigen receptors coordinates the metabolic reprogramming essential for T cell differentiation. \u003cem\u003eNat. Immunol.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e, 500\u0026ndash;508. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/ni.2556\u003c/span\u003e\u003cspan address=\"10.1038/ni.2556\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWaickman, A. T. \u0026amp; Powell, J. D. mTOR, metabolism, and the regulation of T-cell differentiation and function. \u003cem\u003eImmunol. Rev.\u003c/em\u003e \u003cb\u003e249\u003c/b\u003e, 43\u0026ndash;58. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/j.1600-065X.2012.01152.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1600-065X.2012.01152.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePollizzi, K. N. \u0026amp; Powell, J. D. Regulation of T cells by mTOR: the known knowns and the known unknowns. \u003cem\u003eTrends Immunol.\u003c/em\u003e \u003cb\u003e36\u003c/b\u003e, 13\u0026ndash;20. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.it.2014.11.005\u003c/span\u003e\u003cspan address=\"10.1016/j.it.2014.11.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, W. \u0026amp; Zou, W. Amino Acids and Their Transporters in T Cell Immunity and Cancer Therapy. \u003cem\u003eMol. Cell\u003c/em\u003e. \u003cb\u003e80\u003c/b\u003e, 384\u0026ndash;395. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.molcel.2020.09.006\u003c/span\u003e\u003cspan address=\"10.1016/j.molcel.2020.09.006\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlair, M. C., Neinast, M. D. \u0026amp; Arany, Z. Whole-body metabolic fate of branched-chain amino acids. \u003cem\u003eBiochem. J.\u003c/em\u003e \u003cb\u003e478\u003c/b\u003e, 765\u0026ndash;776. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1042/bcj20200686\u003c/span\u003e\u003cspan address=\"10.1042/bcj20200686\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRossi, M. et al. Dietary intake of branched-chain amino acids and colorectal cancer risk. \u003cem\u003eBr. J. Nutr.\u003c/em\u003e \u003cb\u003e126\u003c/b\u003e, 22\u0026ndash;27. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1017/s0007114520003724\u003c/span\u003e\u003cspan address=\"10.1017/s0007114520003724\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu, Z., Qiu, B., Zhou, H., Li, L. \u0026amp; Niu, T. Characterization and application of a lactate and branched chain amino acid metabolism related gene signature in a prognosis risk model for multiple myeloma. \u003cem\u003eCancer Cell Int.\u003c/em\u003e \u003cb\u003e23\u003c/b\u003e, 169. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12935-023-03007-4\u003c/span\u003e\u003cspan address=\"10.1186/s12935-023-03007-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng, J. et al. Integrated single-cell and bulk characterization of branched chain amino acid metabolism-related key gene BCAT1 and association with prognosis and immunogenicity of clear cell renal cell carcinoma. \u003cem\u003eAging\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e, 2715\u0026ndash;2735. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.18632/aging.205506\u003c/span\u003e\u003cspan address=\"10.18632/aging.205506\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, J. Y. et al. GEO data mining and TCGA analysis reveal altered branched chain amino acid metabolism in pancreatic cancer patients. \u003cem\u003eAging\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 11907\u0026ndash;11918. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.18632/aging.202892\u003c/span\u003e\u003cspan address=\"10.18632/aging.202892\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLove, M. I., Huber, W. \u0026amp; Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. \u003cem\u003eGenome biology\u003c/em\u003e 15, 550, (2014). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13059-014-0550-8\u003c/span\u003e\u003cspan address=\"10.1186/s13059-014-0550-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGustavsson, E. K., Zhang, D., Reynolds, R. H., Garcia-Ruiz, S. \u0026amp; Ryten, M. ggtranscript: an R package for the visualization and interpretation of transcript isoforms using ggplot2. \u003cem\u003eBioinf. (Oxford England)\u003c/em\u003e. \u003cb\u003e38\u003c/b\u003e, 3844\u0026ndash;3846. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/bioinformatics/btac409\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/btac409\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGu, Z., Eils, R. \u0026amp; Schlesner, M. Complex heatmaps reveal patterns and correlations in multidimensional genomic data. \u003cem\u003eBioinf. (Oxford England)\u003c/em\u003e. \u003cb\u003e32\u003c/b\u003e, 2847\u0026ndash;2849. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/bioinformatics/btw313\u003c/span\u003e\u003cspan address=\"10.1093/bioinformatics/btw313\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eH\u0026auml;nzelmann, S., Castelo, R. \u0026amp; Guinney, J. GSVA: gene set variation analysis for microarray and RNA-seq data. \u003cem\u003eBMC Bioinform.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/1471-2105-14-7\u003c/span\u003e\u003cspan address=\"10.1186/1471-2105-14-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, T. T. et al. Identification of CDK2-Related Immune Forecast Model and ceRNA in Lung Adenocarcinoma, a Pan-Cancer Analysis. \u003cem\u003eFront. cell. Dev. biology\u003c/em\u003e. \u003cb\u003e9\u003c/b\u003e, 682002. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fcell.2021.682002\u003c/span\u003e\u003cspan address=\"10.3389/fcell.2021.682002\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLangfelder, P. \u0026amp; Horvath, S. WGCNA: an R package for weighted correlation network analysis. \u003cem\u003eBMC Bioinform.\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/1471-2105-9-559\u003c/span\u003e\u003cspan address=\"10.1186/1471-2105-9-559\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, H. \u0026amp; Boutros, P. C. VennDiagram: a package for the generation of highly-customizable Venn and Euler diagrams in R. \u003cem\u003eBMC Bioinform.\u003c/em\u003e \u003cb\u003e12\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/1471-2105-12-35\u003c/span\u003e\u003cspan address=\"10.1186/1471-2105-12-35\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWu, T. et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. \u003cem\u003eInnov. (Cambridge (Mass))\u003c/em\u003e. \u003cb\u003e2\u003c/b\u003e, 100141. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.xinn.2021.100141\u003c/span\u003e\u003cspan address=\"10.1016/j.xinn.2021.100141\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShannon, P. et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. \u003cem\u003eGenome Res.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 2498\u0026ndash;2504. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1101/gr.1239303\u003c/span\u003e\u003cspan address=\"10.1101/gr.1239303\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2003).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLei, J. et al. Clinicopathological characteristics of pheochromocytoma/paraganglioma and screening of prognostic markers. \u003cem\u003eJ. Surg. Oncol.\u003c/em\u003e \u003cb\u003e128\u003c/b\u003e, 510\u0026ndash;518. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/jso.27358\u003c/span\u003e\u003cspan address=\"10.1002/jso.27358\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFang, Y., Huang, S., Han, L., Wang, S. \u0026amp; Xiong, B. Comprehensive Analysis of Peritoneal Metastasis Sequencing Data to Identify LINC00924 as a Prognostic Biomarker in Gastric Cancer. \u003cem\u003eCancer Manage. Res.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 5599\u0026ndash;5611. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.2147/cmar.S318704\u003c/span\u003e\u003cspan address=\"10.2147/cmar.S318704\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEngebretsen, S. \u0026amp; Bohlin, J. Statistical predictions with glmnet. \u003cem\u003eClin. epigenetics\u003c/em\u003e. \u003cb\u003e11\u003c/b\u003e, 123. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13148-019-0730-1\u003c/span\u003e\u003cspan address=\"10.1186/s13148-019-0730-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHeagerty, P. J., Lumley, T. \u0026amp; Pepe, M. S. Time-dependent ROC curves for censored survival data and a diagnostic marker. \u003cem\u003eBiometrics\u003c/em\u003e \u003cb\u003e56\u003c/b\u003e, 337\u0026ndash;344. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/j.0006-341x.2000.00337.x\u003c/span\u003e\u003cspan address=\"10.1111/j.0006-341x.2000.00337.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2000).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, J. et al. A nomogram for predicting prognosis of patients with cervical cerclage. \u003cem\u003eHeliyon\u003c/em\u003e \u003cb\u003e9\u003c/b\u003e, e21147. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.heliyon.2023.e21147\u003c/span\u003e\u003cspan address=\"10.1016/j.heliyon.2023.e21147\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMak, T. K. et al. The cancer-associated fibroblast-related signature predicts prognosis and indicates immune microenvironment infiltration in gastric cancer. \u003cem\u003eFront. Immunol.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 951214. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fimmu.2022.951214\u003c/span\u003e\u003cspan address=\"10.3389/fimmu.2022.951214\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang, Y. et al. A composite score based on immune-related gene prognostic index and m(6)A risk score of head and neck squamous cell carcinoma. \u003cem\u003eFront. Genet.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fgene.2023.1061569\u003c/span\u003e\u003cspan address=\"10.3389/fgene.2023.1061569\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJędroszka, D., Orzechowska, M., Hamouz, R., G\u0026oacute;rniak, K. \u0026amp; Bednarek, A. K. Markers of epithelial-to-mesenchymal transition reflect tumor biology according to patient age and Gleason score in prostate cancer. \u003cem\u003ePloS one\u003c/em\u003e. \u003cb\u003e12\u003c/b\u003e, e0188842. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1371/journal.pone.0188842\u003c/span\u003e\u003cspan address=\"10.1371/journal.pone.0188842\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRobles-Jimenez, L. E. et al. Worldwide Traceability of Antibiotic Residues from Livestock in Wastewater and Soil: A Systematic Review. \u003cem\u003eAnimals: open. access. J. MDPI\u003c/em\u003e. \u003cb\u003e12\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/ani12010060\u003c/span\u003e\u003cspan address=\"10.3390/ani12010060\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheng, J., Chen, F. \u0026amp; Cheng, Y. Construction and Evaluation of a Risk Score Model for Lymph Node Metastasis-Associated Circadian Clock Genes in Esophageal Squamous Carcinoma. \u003cem\u003eCells\u003c/em\u003e 11, (2022). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/cells11213432\u003c/span\u003e\u003cspan address=\"10.3390/cells11213432\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMayakonda, A., Lin, D. C., Assenov, Y., Plass, C. \u0026amp; Koeffler, H. P. Maftools: efficient and comprehensive analysis of somatic variants in cancer. \u003cem\u003eGenome Res.\u003c/em\u003e \u003cb\u003e28\u003c/b\u003e, 1747\u0026ndash;1756. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1101/gr.239244.118\u003c/span\u003e\u003cspan address=\"10.1101/gr.239244.118\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaeser, D., Gruener, R. F. \u0026amp; Huang, R. S. oncoPredict: an R package for predicting in vivo or cancer patient drug response and biomarkers from cell line screening data. \u003cem\u003eBrief. Bioinform.\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/bib/bbab260\u003c/span\u003e\u003cspan address=\"10.1093/bib/bbab260\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHao, Y. et al. Integrated analysis of multimodal single-cell data. \u003cem\u003eCell\u003c/em\u003e \u003cb\u003e184\u003c/b\u003e, 3573\u0026ndash;3587e3529. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.cell.2021.04.048\u003c/span\u003e\u003cspan address=\"10.1016/j.cell.2021.04.048\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHao, X. et al. Inhibition of APOC1 promotes the transformation of M2 into M1 macrophages via the ferroptosis pathway and enhances anti-PD1 immunotherapy in hepatocellular carcinoma based on single-cell RNA sequencing. \u003cem\u003eRedox Biol.\u003c/em\u003e \u003cb\u003e56\u003c/b\u003e, 102463. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.redox.2022.102463\u003c/span\u003e\u003cspan address=\"10.1016/j.redox.2022.102463\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQiu, X. et al. Reversed graph embedding resolves complex single-cell trajectories. \u003cem\u003eNat. Methods\u003c/em\u003e. \u003cb\u003e14\u003c/b\u003e, 979\u0026ndash;982. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/nmeth.4402\u003c/span\u003e\u003cspan address=\"10.1038/nmeth.4402\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, E., Ji, B., Jin, K. \u0026amp; Chen, Y. Branched-chain amino acids catabolism and cancer progression: focus on therapeutic interventions. \u003cem\u003eFront. Oncol.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 1220638. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2023.1220638\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2023.1220638\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBonvini, A., Coqueiro, A. Y., Tirapegui, J., Calder, P. C. \u0026amp; Rogero, M. M. Immunomodulatory role of branched-chain amino acids. \u003cem\u003eNutr. Rev.\u003c/em\u003e \u003cb\u003e76\u003c/b\u003e, 840\u0026ndash;856. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/nutrit/nuy037\u003c/span\u003e\u003cspan address=\"10.1093/nutrit/nuy037\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSivanand, S. \u0026amp; Vander Heiden, M. G. Emerging Roles for Branched-Chain Amino Acid Metabolism in Cancer. \u003cem\u003eCancer cell.\u003c/em\u003e \u003cb\u003e37\u003c/b\u003e, 147\u0026ndash;156. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.ccell.2019.12.011\u003c/span\u003e\u003cspan address=\"10.1016/j.ccell.2019.12.011\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePegg, A. E. \u0026amp; Michael, A. J. Spermine synthase. \u003cem\u003eCell. Mol. Life Sci.\u003c/em\u003e \u003cb\u003e67\u003c/b\u003e, 113\u0026ndash;121. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00018-009-0165-5\u003c/span\u003e\u003cspan address=\"10.1007/s00018-009-0165-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXiang, L., Piao, L., Wang, D. \u0026amp; Qi, L. F. Overexpression of SMS in the tumor microenvironment is associated with immunosuppression in hepatocellular carcinoma. \u003cem\u003eFront. Immunol.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 974241. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fimmu.2022.974241\u003c/span\u003e\u003cspan address=\"10.3389/fimmu.2022.974241\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFahrmann, J. F. et al. Association Between Plasma Diacetylspermine and Tumor Spermine Synthase With Outcome in Triple-Negative Breast Cancer. \u003cem\u003eJ. Natl Cancer Inst.\u003c/em\u003e \u003cb\u003e112\u003c/b\u003e, 607\u0026ndash;616. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/jnci/djz182\u003c/span\u003e\u003cspan address=\"10.1093/jnci/djz182\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo, Y. et al. Spermine synthase and MYC cooperate to maintain colorectal cancer cell survival by repressing Bim expression. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cb\u003e11\u003c/b\u003e, 3243. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41467-020-17067-x\u003c/span\u003e\u003cspan address=\"10.1038/s41467-020-17067-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePan, X., Xue, L. \u0026amp; Sun, Y. Spermine synthase (SMS) serves as a prognostic biomarker in head and neck squamous cell carcinoma: a bioinformatics analysis. \u003cem\u003eAnnals translational Med.\u003c/em\u003e \u003cb\u003e10\u003c/b\u003e, 1213. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21037/atm-22-5014\u003c/span\u003e\u003cspan address=\"10.21037/atm-22-5014\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNovita Sari, I. et al. Metabolism and function of polyamines in cancer progression. \u003cem\u003eCancer Lett.\u003c/em\u003e \u003cb\u003e519\u003c/b\u003e, 91\u0026ndash;104. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.canlet.2021.06.020\u003c/span\u003e\u003cspan address=\"10.1016/j.canlet.2021.06.020\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJung, M. K., Okekunle, A. P., Lee, J. E., Sung, M. K. \u0026amp; Lim, Y. J. Role of Branched-chain Amino Acid Metabolism in Tumor Development and Progression. \u003cem\u003eJ. cancer Prev.\u003c/em\u003e \u003cb\u003e26\u003c/b\u003e, 237\u0026ndash;243. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.15430/jcp.2021.26.4.237\u003c/span\u003e\u003cspan address=\"10.15430/jcp.2021.26.4.237\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDing, C., Fan, X. \u0026amp; Wu, G. Peroxiredoxin 1 - an antioxidant enzyme in cancer. \u003cem\u003eJ. Cell. Mol. Med.\u003c/em\u003e \u003cb\u003e21\u003c/b\u003e, 193\u0026ndash;202. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/jcmm.12955\u003c/span\u003e\u003cspan address=\"10.1111/jcmm.12955\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYin, K. et al. Using weighted gene co-expression network analysis to identify key modules and hub genes in tongue squamous cell carcinoma. \u003cem\u003eMedicine\u003c/em\u003e \u003cb\u003e98\u003c/b\u003e, e17100. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1097/md.0000000000017100\u003c/span\u003e\u003cspan address=\"10.1097/md.0000000000017100\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, L. et al. A Novel Ferroptosis-Related Gene Signature to Predict Prognosis in Patients with Head and Neck Squamous Cell Carcinoma. \u003cem\u003eDisease markers\u003c/em\u003e 5759927, (2021). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1155/2021/5759927\u003c/span\u003e\u003cspan address=\"10.1155/2021/5759927\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLeopoldino, A. M. et al. SET protein accumulates in HNSCC and contributes to cell survival: antioxidant defense, Akt phosphorylation and AVOs acidification. \u003cem\u003eOral Oncol.\u003c/em\u003e \u003cb\u003e48\u003c/b\u003e, 1106\u0026ndash;1113. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.oraloncology.2012.05.014\u003c/span\u003e\u003cspan address=\"10.1016/j.oraloncology.2012.05.014\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDjukic, T. et al. Upregulated glutathione transferase omega-1 correlates with progression of urinary bladder carcinoma. \u003cem\u003eRedox report: Commun. free radical Res.\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e, 486\u0026ndash;492. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/13510002.2017.1299909\u003c/span\u003e\u003cspan address=\"10.1080/13510002.2017.1299909\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, K., Zhang, F. L. \u0026amp; Jia, W. Glutathione S\u0026ndash;transferase ω 1 promotes the proliferation, migration and invasion, and inhibits the apoptosis of non\u0026ndash;small cell lung cancer cells, via the JAK/STAT3 signaling pathway. \u003cem\u003eMol. Med. Rep.\u003c/em\u003e \u003cb\u003e23\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3892/mmr.2020.11709\u003c/span\u003e\u003cspan address=\"10.3892/mmr.2020.11709\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, Y. et al. Identification of glutathione S-transferase omega 1 (GSTO1) protein as a novel tumor-associated antigen and its autoantibody in human esophageal squamous cell carcinoma. \u003cem\u003eTumour biology: J. Int. Soc. Oncodevelopmental Biology Med.\u003c/em\u003e \u003cb\u003e35\u003c/b\u003e, 10871\u0026ndash;10877. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s13277-014-2394-y\u003c/span\u003e\u003cspan address=\"10.1007/s13277-014-2394-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhenyukh, O. et al. Branched-chain amino acids promote endothelial dysfunction through increased reactive oxygen species generation and inflammation. \u003cem\u003eJ. Cell. Mol. Med.\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e, 4948\u0026ndash;4962. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1111/jcmm.13759\u003c/span\u003e\u003cspan address=\"10.1111/jcmm.13759\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZavialov, A. V. \u0026amp; Engstr\u0026ouml;m, A. Human ADA2 belongs to a new family of growth factors with adenosine deaminase activity. \u003cem\u003eBiochem. J.\u003c/em\u003e \u003cb\u003e391\u003c/b\u003e, 51\u0026ndash;57. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1042/bj20050683\u003c/span\u003e\u003cspan address=\"10.1042/bj20050683\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2005).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKaljas, Y. et al. Human adenosine deaminases ADA1 and ADA2 bind to different subsets of immune cells. \u003cem\u003eCell. Mol. Life Sci.\u003c/em\u003e \u003cb\u003e74\u003c/b\u003e, 555\u0026ndash;570. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s00018-016-2357-0\u003c/span\u003e\u003cspan address=\"10.1007/s00018-016-2357-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYe, Z., Wang, S., Zhang, C. \u0026amp; Zhao, Y. Coordinated Modulation of Energy Metabolism and Inflammation by Branched-Chain Amino Acids and Fatty Acids. \u003cem\u003eFront. Endocrinol.\u003c/em\u003e \u003cb\u003e11\u003c/b\u003e, 617. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fendo.2020.00617\u003c/span\u003e\u003cspan address=\"10.3389/fendo.2020.00617\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeng, J. et al. Emerging roles and the regulation of aerobic glycolysis in hepatocellular carcinoma. \u003cem\u003eJ. experimental Clin. cancer research: CR\u003c/em\u003e. \u003cb\u003e39\u003c/b\u003e, 126. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s13046-020-01629-4\u003c/span\u003e\u003cspan address=\"10.1186/s13046-020-01629-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, Z. et al. HOXA1 promotes aerobic glycolysis and cancer progression in cervical cancer. \u003cem\u003eCell. Signal.\u003c/em\u003e \u003cb\u003e109\u003c/b\u003e, 110747. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.cellsig.2023.110747\u003c/span\u003e\u003cspan address=\"10.1016/j.cellsig.2023.110747\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlfarouk, K. O. et al. The Pentose Phosphate Pathway Dynamics in Cancer and Its Dependency on Intracellular pH. \u003cem\u003eMetabolites\u003c/em\u003e 10, (2020). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/metabo10070285\u003c/span\u003e\u003cspan address=\"10.3390/metabo10070285\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChacon-Barahona, J. A., MacKeigan, J. P. \u0026amp; Lanning, N. J. Unique Metabolic Contexts Sensitize Cancer Cells and Discriminate between Glycolytic Tumor Types. \u003cem\u003eCancers\u003c/em\u003e \u003cb\u003e15\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3390/cancers15041158\u003c/span\u003e\u003cspan address=\"10.3390/cancers15041158\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXu, X. et al. Association of Germline Variants in Natural Killer Cells With Tumor Immune Microenvironment Subtypes, Tumor-Infiltrating Lymphocytes, Immunotherapy Response, Clinical Outcomes, and Cancer Risk. \u003cem\u003eJAMA Netw. open.\u003c/em\u003e \u003cb\u003e2\u003c/b\u003e, e199292. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1001/jamanetworkopen.2019.9292\u003c/span\u003e\u003cspan address=\"10.1001/jamanetworkopen.2019.9292\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWetzel, T. J., Erfan, S. C., Figueroa, L. D., Wheeler, L. M. \u0026amp; Ananieva, E. A. Crosstalk between arginine, glutamine, and the branched chain amino acid metabolism in the tumor microenvironment. \u003cem\u003eFront. Oncol.\u003c/em\u003e \u003cb\u003e13\u003c/b\u003e, 1186539. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3389/fonc.2023.1186539\u003c/span\u003e\u003cspan address=\"10.3389/fonc.2023.1186539\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWulff, S., Pries, R., B\u0026ouml;rngen, K., Trenkle, T. \u0026amp; Wollenberg, B. Decreased levels of circulating regulatory NK cells in patients with head and neck cancer throughout all tumor stages. \u003cem\u003eAnticancer Res.\u003c/em\u003e \u003cb\u003e29\u003c/b\u003e, 3053\u0026ndash;3057 (2009).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCooper, M. A. et al. Human natural killer cells: a unique innate immunoregulatory role for the CD56(bright) subset. \u003cem\u003eBlood\u003c/em\u003e \u003cb\u003e97\u003c/b\u003e, 3146\u0026ndash;3151. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1182/blood.v97.10.3146\u003c/span\u003e\u003cspan address=\"10.1182/blood.v97.10.3146\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2001).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLoos, J. et al. Functional characteristics of Th1, Th17, and ex-Th17 cells in EAE revealed by intravital two-photon microscopy. \u003cem\u003eJ. Neuroinflamm.\u003c/em\u003e \u003cb\u003e17\u003c/b\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12974-020-02021-x\u003c/span\u003e\u003cspan address=\"10.1186/s12974-020-02021-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeng, S., Chen, X. M., Wang, J. F. \u0026amp; Xu, X. Q. Th17 cells associated cytokines and cancer. \u003cem\u003eEur. Rev. Med. Pharmacol. Sci.\u003c/em\u003e \u003cb\u003e20\u003c/b\u003e, 4032\u0026ndash;4040 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYoung, M. R., Levingston, C. A. \u0026amp; Johnson, S. D. Treatment to sustain a Th17-type phenotype to prevent skewing toward Treg and to limit premalignant lesion progression to cancer. \u003cem\u003eInt. J. Cancer\u003c/em\u003e. \u003cb\u003e138\u003c/b\u003e, 2487\u0026ndash;2498. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/ijc.29989\u003c/span\u003e\u003cspan address=\"10.1002/ijc.29989\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCai, Z., Tang, B., Chen, L. \u0026amp; Lei, W. Mast cell marker gene signature in head and neck squamous cell carcinoma. \u003cem\u003eBMC cancer\u003c/em\u003e. \u003cb\u003e22\u003c/b\u003e, 577. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12885-022-09673-3\u003c/span\u003e\u003cspan address=\"10.1186/s12885-022-09673-3\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, Z. et al. Genetic and molecular characterization of metabolic pathway-based clusters in esophageal squamous cell carcinoma. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cb\u003e14\u003c/b\u003e, 6200. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41598-024-56391-w\u003c/span\u003e\u003cspan address=\"10.1038/s41598-024-56391-w\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuigay, J. et al. Cetuximab, docetaxel, and cisplatin versus platinum, fluorouracil, and cetuximab as first-line treatment in patients with recurrent or metastatic head and neck squamous-cell carcinoma (GORTEC 2014-01 TPExtreme): a multicentre, open-label, randomised, phase 2 trial. \u003cem\u003eLancet Oncol.\u003c/em\u003e \u003cb\u003e22\u003c/b\u003e, 463\u0026ndash;475. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/s1470-2045(20)30755-5\u003c/span\u003e\u003cspan address=\"10.1016/s1470-2045(20)30755-5\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eForastiere, A. A. Use of paclitaxel (Taxol) in squamous cell carcinoma of the head and neck. \u003cem\u003eSemin. Oncol.\u003c/em\u003e \u003cb\u003e20\u003c/b\u003e, 56\u0026ndash;60 (1993).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, S., Lian, M., Han, B., Fang, J. \u0026amp; Wang, Z. Single-cell integrated transcriptomics reveals the role of keratinocytes in head and neck squamous cell carcinoma. \u003cem\u003eJ. Appl. Genet.\u003c/em\u003e \u003cb\u003e65\u003c/b\u003e, 727\u0026ndash;745. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1007/s13353-024-00842-7\u003c/span\u003e\u003cspan address=\"10.1007/s13353-024-00842-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSun, M. et al. Cuproptosis-related lncRNA JPX regulates malignant cell behavior and epithelial-immune interaction in head and neck squamous cell carcinoma via miR-193b-3p/PLAU axis. \u003cem\u003eInt. J. Oral Sci.\u003c/em\u003e \u003cb\u003e16\u003c/b\u003e, 63. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41368-024-00314-y\u003c/span\u003e\u003cspan address=\"10.1038/s41368-024-00314-y\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLai, Y. J. et al. CSC-3436 inhibits TWIST-induced epithelial-mesenchymal transition via the suppression of Twist/Bmi1/Akt pathway in head and neck squamous cell carcinoma. \u003cem\u003eJ. Cell. Physiol.\u003c/em\u003e \u003cb\u003e234\u003c/b\u003e, 9118\u0026ndash;9129. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/jcp.27589\u003c/span\u003e\u003cspan address=\"10.1002/jcp.27589\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao, M. et al. Targeting of EZH2 inhibits epithelial\u0026ndash;mesenchymal transition in head and neck squamous cell carcinoma via regulating the STAT3/VEGFR2 axis. \u003cb\u003e55\u003c/b\u003e, 1165\u0026ndash;1175, (2019). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.3892/ijo.2019.4880\u003c/span\u003e\u003cspan address=\"10.3892/ijo.2019.4880\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\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":"Head and neck squamous cell carcinoma, branched chain amino acids metabolism, Single-cell RNA sequencing, Epithelial cells, Risk model","lastPublishedDoi":"10.21203/rs.3.rs-6030799/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6030799/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSeveral studies have demonstrated that impaired metabolism of branched chain amino acids (BCAAs) is related to cancer progression. However, the specific mechanisms underlying BCAA metabolism in head and neck squamous cell carcinoma (HNSCC) remain to be explored. The aim of this study was to identify prognostic genes associated with BCAA metabolism in HNSCC and to elucidate their functional mechanisms.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe HNSCC related datasets (TCGA-HNSCC, GSE65858 and GSE140042) were enrolled in this study. Candidate genes were acquired by overlapping differentially expressed genes form differential expression analysis and key module genes connected with BCAA-metabolism related genes (BCAA-MRGs) scores from weighted gene co-expression network analysis. Subsequently, prognostic genes were obtained to construct the risk model through univariate Cox regression analysis, proportional hazards hypothesis test, and least absolute shrinkage and selection operator regression analysis selected in sequence. Afterwards, independent prognostic analysis, enrichment analysis and immune microenvironment analysis were performed. Furthermore, the expression changes of prognostic genes at the cellular level were assessed through single-cell RNA sequencing (scRNA-seq) data analysis and pseudo-time analysis. Additionally, RT-qPCR was used to confirm the expression levels of prognostic genes in HNSCC tissues.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSMS, PRDX6, GSTO1, and ADA were determined as prognostic genes to create the risk model. The HNSCC samples were divided into high-risk group (HRG) and low-risk group (LRG), with LRG demonstrating significantly higher survival rates compared to the HRG. Furthermore, the nomogram model constructed using risk score and age had an excellent predictive ability for HNSCC patients. Enrichment analysis revealed that ‘pentose phosphate pathway’ and ‘fructose and mannose metabolism’ were significantly associated with HNSCC progression. At the same time, we also found that the level of infiltration of 20 immune cells (plasmacytoid dendritic cells, mast cells, and T follicle helper cells) and the expression of 10 immune checkpoints (CD276, CD27, and CD40) differed between the HRG and the LRG. Additionally, epithelial cells were selected as key cells due to higher expression of prognostic genes. Importantly, the trend of prognostic gene expression varied with different stages of cell differentiation. Through RT-qPCR experiment, SMS, GSTO1, and ADA all expressed highly in the tumor group, but PRDX6 had not remarkably difference between tumor and normal groups.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn summary, we pinpointed four genes-SMS, PRDX6, GSTO1, and ADA-linked to the prognosis of HNSCC within the context of BCAA metabolism. Subsequently, we developed a risk model. This model offers a novel reference for prognostic assessment and treatment strategies tailored to HNSCC patients.\u003c/p\u003e","manuscriptTitle":"Integration RNA bulk and single cell RNA sequencing to explore the change of BCAA metabolism-related immune microenvironment and construct prognostic signature in HNSCC","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-25 09:37:43","doi":"10.21203/rs.3.rs-6030799/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":"7d57d2be-ee77-41cf-99fd-6625c2b4c615","owner":[],"postedDate":"June 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":48696188,"name":"Biological sciences/Cancer"},{"id":48696189,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":48696190,"name":"Biological sciences/Drug discovery"},{"id":48696191,"name":"Biological sciences/Immunology"},{"id":48696192,"name":"Health sciences/Biomarkers"}],"tags":[],"updatedAt":"2025-07-09T12:24:02+00:00","versionOfRecord":[],"versionCreatedAt":"2025-06-25 09:37:43","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6030799","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6030799","identity":"rs-6030799","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}