Dissecting T Cell Exhaustion in Non-Small Cell Lung Cancer: Single-Cell and Spatial Transcriptomics Reveal Prognostic Signatures and Therapeutic Implications

Dissecting T Cell Exhaustion in Non-Small Cell Lung Cancer: Single-Cell and Spatial Transcriptomics Reveal Prognostic Signatures and Therapeutic Implications | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Dissecting T Cell Exhaustion in Non-Small Cell Lung Cancer: Single-Cell and Spatial Transcriptomics Reveal Prognostic Signatures and Therapeutic Implications Qingwen Hu, Haiqing Chen, Haotian Lai, Lai Jiang, Xuancheng Zhou, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7434355/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: Non-small cell lung cancer (NSCLC) represents 85% of lung cancers and remains a leading cause of cancer death. Immunotherapy advancements have improved treatment, but many patients develop resistance due to T cell exhaustion. Understanding this mechanism, aided by single-cell RNA sequencing, is vital for creating personalized therapies. Methods: Single-cell RNA sequencing (scRNA-seq) and bulk RNA-seq data from NSCLC patients and normal tissues were collected from multiple databases. Batch effects were corrected, and scRNA-seq data were processed using Seurat for dimensionality reduction and clustering. Exhausted T cell subpopulations were identified, and transcriptional and spatial analyses were conducted using SCENIC and pseudotime analysis. Additionally, a 20-T-ExhauRs prognostic model was developed using machine learning algorithms, and immune infiltration and drug sensitivity analyses were performed. Statistical analyses were conducted using R and Python software. Results: The study identified exhausted T cell subpopulations in NSCLC using dimensionality reduction and clustering, revealing 25 subpopulations and significant differences between normal and NSCLC groups. Pseudotime and transcription factor analysis showed the evolution of exhausted T cell subpopulations. Spatial transcriptomics and metabolic pathway enrichment revealed heterogeneity in the tumor microenvironment. The 20-T-ExhauRs prognostic model was developed using machine learning and demonstrated strong survival prediction accuracy. Immune infiltration analysis revealed weaker immune responses in high-risk groups, while drug sensitivity analysis indicated reduced effectiveness of certain chemotherapies. The study offers insights into immune regulation, tumor progression, and therapeutic strategies for NSCLC. Conclusion: This study identified exhausted T cell subpopulations in NSCLC, revealing their roles in tumor progression. The 20-T-ExhauRs model accurately predicted survival outcomes. Spatial transcriptomics and immune infiltration analyses highlighted tumor heterogeneity, suggesting potential therapeutic strategies to improve NSCLC treatment and patient prognosis. NSCLC T cell exhaustion scRNA-seq Spatial transcriptomics and Prognostic models Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 1. Introduction NSCLC, representing 85% of all pulmonary malignancies [1], continues to impose substantial global cancer mortality despite therapeutic advances [2]. Immune checkpoint inhibitors (ICIs) have transformed treatment strategies; however, 60–70% of patients still experience limited efficacy or develop resistance [3]. Increasing evidence points to T cell exhaustion (Tex)—a dysfunctional state marked by progressive loss of effector functions and sustained expression of inhibitory receptors—as a central mechanism of immune escape and treatment failure [4]. Tex arises from prolonged antigen exposure in the tumor microenvironment (TME), highlighting the need to elucidate NSCLC-specific exhaustion pathways to inform therapeutic innovation. The NSCLC TME promotes Tex development through complex interactions among tumor cells, immunosuppressive factors, and metabolic stressors [5]. Although prior studies have linked Tex to tumor progression, the functional heterogeneity and spatial distribution of Tex subsets remain inadequately defined [6]. Advances in single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics now offer powerful tools for profiling cellular states with spatial precision [7] , enabling detailed exploration of Tex diversity and its clinical significance. In this multicenter study, we applied an integrative multi-omics approach to systematically investigate Tex dynamics in NSCLC. We used and analyzed scRNA-seq and spatial transcriptome data to identify distinct Tex subpopulations with distinct clinical associations. Pseudotime analysis traced exhaustion trajectories, while ligand-receptor interaction networks revealed key microenvironmental contributors to Tex induction. To facilitate clinical translation, we constructed a 20-gene Tex signature (20-T-ExhauRs) using an ensemble machine learning model incorporating ten algorithms, demonstrating robust prognostic value. Additional analyses linked Tex states with patterns of immunotherapy resistance and potential therapeutic vulnerabilities. Overall, this study aims to dissect the phenotypic and functional landscape of exhausted T cells in NSCLC and clarify their role within the TME, offering a theoretical basis for personalized immunotherapy. By advancing our understanding of Tex, we strive to support the development of precision strategies for NSCLC treatment. 2. Methods and Materials 2.1 Data acquisition and preprocessing scRNA-seq data from NSCLC patients (GSM5938737: NSCLC1, GSM5938738: NSCLC2) and normal lung tissues (GSM5938739: CT1, GSM5938740: CT2) were obtained from the GSE198099 dataset in the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). Additionally, bulk RNA-seq data from 272 NSCLC patients in GSE30219 and 52 patients in GSE29016 were analyzed. Genomic profiles from 504 lung squamous cell carcinoma (LUSC) and 522 lung adenocarcinoma (LUAD) cases were retrieved from The Cancer Genome Atlas (TCGA) via the GDC portal (https://portal.gdc.cancer.gov/). Spatial transcriptomics data were sourced from the BioStudies database (https://www.ebi.ac.uk/biostudies/) [8]. To minimize technical variation and ensure cross-platform comparability, batch effects were corrected using the "ComBat" algorithm, implemented through the "limma" (v3.50.3) [9] and "sva" (v3.48.0) R packages. For downstream analysis, TCGA-LUSC and TCGA-LUAD cohorts were designated as the training set, while GSE30219 and GSE29016 served as independent validation datasets. 2.2 scRNA-seq data were processed to obtain target cell populations scRNA-seq data were processed using the Seurat package [10]. We control for quality to be less than 200 genes or more than 4000 genes, or mitochondrial gene content >20% of the cells were excluded. Highly variable genes were identified via the FindVariableFeatures() function [10] and visualized using VariableFeaturePlot() [11]. Following normalization and scaling using ScaleData(), principal component analysis (PCA) was employed for linear dimensionality reduction. Non-linear embedding techniques, including t-distributed stochastic neighbor embedding (t-SNE) and uniform manifold approximation and projection (UMAP), were applied using RunTSNE() and RunUMAP() respectively. Cell clustering was performed through neighborhood graph construction (FindNeighbors()) followed by community detection (FindClusters()). Cell type annotation was achieved through a two-pronged strategy: (1) automated prediction using the singleR package (v2.4.0) [12]; and (2) manual confirmation based on canonical marker genes, referencing the CellMarker database [13] and previously reported markers. 2.3 Exhausted T Cell Subpopulation Analysis To characterize exhausted T cell subtypes, integrated computational analyses were conducted. Seurat (v4.3.0) [10] facilitated dimensionality reduction, clustering, and phenotype labeling. Batch effects were mitigated using the Harmony algorithm [14] , followed by UMAP for spatial representation. Intercellular communication networks were inferred using CellChat (v1.6.1) [15] via a three-step pipeline: ligand-receptor interaction prediction, signaling pathway enrichment analysis, and network topology assessment. Network visualization employed netVisual_circle, while nodal centrality was evaluated using netAnalysis_signalingRole_network. To explore regulatory programs, transcription factor activity and regulon dynamics were inferred using SCENIC (v1.3.1) [16] , enabling characterization of cell-type-specific transcriptional circuits in exhausted T cells. Pseudotime analysis was performed with Monocle2 (v2.22.0) [17] , reconstructing the differentiation continuum and gene expression trajectories associated with T cell exhaustion. 2.4 Spatial Transcriptomics Data Analysis Spatial transcriptomics analysis was conducted using Seurat (v.4.3.0) [10]. To identify highly variable genes, we used the SCTransform algorithm [18] for normalization and variance stabilization, in order to identify highly variable genes. Dimensionality reduction via UMAP and visualization using SpatialFeaturePlot enabled spatial mapping of cell subpopulations within intact tissue structures, revealing spatial heterogeneity and functional localization. Comparative metabolic profiling between tumor and adjacent normal tissues was performed using “scMetabolism” (v0.2.1) [19] , which incorporates 167 curated pathways (85 KEGG, 82 REACTOME) via the VISION algorithm for activity scoring. Exhaustion-associated metabolic features were further quantified using AddModuleScore, AUCell, and singscore algorithms [20] based on a 92-gene exhaustion signature, allowing evaluation of glucose metabolism, nucleotide biosynthesis, and lipid metabolism within immune niches. Additional preprocessing and downstream analyses were carried out using “Scanpy” (v1.4.5) [21] , encompassing spatial clustering, trajectory inference, and differential gene expression analysis. The stLearn framework (https://github.com/BiomedicalMachineLearning/stLearn) enhanced spatial feature extraction by integrating PCA-based reduction with histological image segmentation. Spatial Morphological Eigenvector (SME) adjustment and k-nearest neighbor (KNN) graph construction preceded cluster resolution optimization via the Louvain-Leiden algorithm. Trajectory inference was again performed using Monocle2 (v2.22.0) [17] , focusing on spatial clusters 0 and 5 to uncover pseudotemporal transitions. Key transitional markers were identified after ribosomal gene filtering, and trajectory-specific genes were ranked. Functional enrichment of the top 30 genes was conducted through Metascape (https://metascape.org/gp/index.html) using hypergeometric testing and Benjamini-Hochberg correction. 2.5 Construction of the 20-T-ExhauRs Model, Nomogram Development, and Independent Prognostic Analysis To construct the 20-T-ExhauRs prognostic signature, we implemented a comprehensive machine learning framework integrating 10 algorithms, yielding 101 combinatorial models. Prognostically relevant genes were initially identified via univariate Cox regression with Benjamini-Hochberg adjustment (adjusted p < 0.05). Model performance was comparatively assessed across training and validation datasets, with the CoxBoost-Enet hybrid model (Enet α = 0.6) emerging as optimal based on concordance index metrics. Individual patient risk scores were calculated using the formula: Risk Score = ExpressionmRNA1 × CoefmRNA1 + ExpressionmRNA2 × CoefmRNA2 + … + ExpressionmRNAn × CoefmRNAn [22]. Survival analyses were performed with the use of the "Survival" (v3.5-5) and "survminer" (v0.4.9) R packages. The Kaplan-Meier curve was used to compare the survival outcomes between the high and low risk groups, and the statistical significance was evaluated by the log-rank test p value [23]. A prognostic nomogram was constructed to estimate 1-, 3-, and 5-year survival probabilities, and its predictive performance was evaluated using calibration curves. Univariate and multivariate Cox regression analyses were also performed, with results visualized in forest plots to assess whether the risk score and nomogram served as independent prognostic indicators [24]. 2.6 Immune Microenvironment Analysis The CIBERSORT algorithm [25] was applied to quantify immune cell infiltration, retaining results with Benjamini-Hochberg-adjusted p-values < 0.05 [26]. Expression matrices were normalized and reshaped using the "limma" (v3.50.3) [9] and "reshape2" (v1.4.1) packages. We further analyzed immune heterogeneity by means of single-sample gene-set enrichment analysis (ssGSEA), calculating enrichment scores for 28 immune-cell subsets and 12 immune-related pathways across risk groups. Group comparisons of immune infiltration levels and functional pathway activities were visualized using the "ggpubr" package (v0.6.0), applying Wilcoxon rank-sum tests (p < 0.05). Additionally, a three-dimensional immune landscape analysis highlighted spatial differences in dendritic cells, neutrophils, and checkpoint-related lymphocytes between risk categories. 2.7 Therapeutic Response Profiling To evaluate therapeutic susceptibility, pharmacogenomic correlations between 17 programmed cell death-related genes (17-PCDRGs) and drug sensitivity were analyzed using data from the Genomics of Drug Sensitivity in Cancer (GDSC) database [27] . We used the GDSC database to explore the relationship between the expression of 17-PCDRGs and the half-maximal inhibitory concentration (IC50) of anti-tumor drugs. Drug sensitivity was predicted using the "OncoPredict" R package, with a filtering condition set to P<0.001 [28]. 2.8 Statistical Analysis All statistical analyses were conducted using R (v4.4.1 and v4.1.3) and Python (v3.9). Correlation analyses were conducted using Spearman's rank method. One-way ANOVA and t-tests were employed to assess differences in 20-T-ExhauRs scores, pathway enrichment levels, immune cell infiltration, and functional pathway scores. Statistical significance was defined as p < 0.05 and false discovery rate (FDR) q < 0.05. 3. Results We developed a flow chart to reveal the ideas of this study (Fig.1). 3.1 Characterization of T Cell Exhaustion Subpopulations Unsupervised clustering of NSCLC single-cell transcriptomic profiles, following nonlinear dimensionality reduction, identified 25 distinct cell clusters, visualized via t-SNE and UMAP (Fig.2A). Cluster annotation based on canonical lineage markers (Fig.2B) and subsequent cluster refinement yielded 10 major cell types: T cells, mononuclear phagocytes, B cells, macrophages, epithelial cells, plasma cells, dendritic cells, endothelial cells, mast cells, and smooth muscle cells (Fig.2C). Ligand-receptor interaction analysis indicated extensive intercellular communication across these cell types (Fig.2D). Comparison between tumor and normal lung tissues revealed significant differences in T cell infiltration, with a notable increase in T cell abundance within NSCLC samples (p < 0.001; Fig.2E). Further subclustering of T cells distinguished two functionally distinct phenotypes: activated T cells (CD69⁺CD28⁺) and exhausted T cells (CD38⁺PDCD1⁺). Quantitative analysis demonstrated a twofold enrichment of exhausted T cells in NSCLC tissues compared to controls (p < 0.001; Fig.2F). 3.2 Developmental Trajectory and Regulatory Network Analysis of Exhausted T Cell Subsets To further explore the heterogeneity of exhausted T cells in NSCLC, we applied non-negative matrix factorization (NMF), coupled with pseudotime analysis, intercellular communication modeling, and transcription factor profiling. NMF identified seven transcriptionally distinct exhausted T cell clusters (Fig.3A). Pseudotime trajectory reconstruction showed a gradual increase in exhaustion markers such as PDCD1, LAG3, and CTLA4, which peaked in the intermediate-to-late stages of differentiation (Pearson r > 0.7, p < 0.001; Fig.3B). Differentiation trajectories converged toward Cluster 4, representing the terminally exhausted state (Fig.3C). We defined the exhausted T cell subsets based on signature genes as follows: TNFRSF18⁺, GZMB⁺, CXCL13⁺, DUSP4⁺, IFNG⁺, and non-aggregated exhausted T cells (Fig.3D). Notably, endothelial cells exhibited high communication frequency with exhausted T cell subsets, especially with the IFNG⁺ cluster, along with strong interactions among exhausted T cell subsets themselves (Fig.3E). Signaling pathway analysis revealed active participation of several ligand-receptor axes—including MIF, CXCL, CCL, GALECTIN, PARs, VISFATIN, MK, VEGF, ANGPTL, IFN-II, CD137, IL16, CALCR, CD40, SEMA3, and FASLG—in both incoming and outgoing signaling among the exhausted T cell subsets (Fig.3F). Regulon activity profiling further uncovered key transcriptional regulators with prominent expression across subsets, including STAT1, REL, IRF1, NR3C1, ATF4, CREM, ELF1, JUNB, FOS, FOSB, JUND, JUN, YY1, CEBPB, ZNF683, and ATF3 (Fig.3G), suggesting their involvement in the establishment and maintenance of T cell exhaustion phenotypes. 3.3 Transcriptomic Heterogeneity of T Cell Exhaustion Markers in NSCLC Microenvironments Transcriptomic comparisons of normal and NSCLC tissues demonstrated distinct cellular compositions and exhaustion-related gene profiles. Integrated scoring (AUCell, AddModuleScore) and heatmap visualization delineated phenotypic variation in stromal subsets—smooth muscle, mast, and endothelial cells—across tumor microenvironments (Fig.4A). T cell functional analysis showed marked metabolic alterations and enhanced exhaustion pathway activity in NSCLC (P<0.001; Fig.4B). t-SNE dimensionality reduction identified distinct clusters with exhaustion signatures, revealing spatial segregation of high-scoring NSCLC cells versus normal counterparts (Fig.4C). Microenvironment-specific activation was evident in violin plots: immune populations (T cells, macrophages, monocytes) displayed elevated exhaustion scores in NSCLC (Δscore >2), whereas structural cells (epithelial, endothelial) maintained normal tissue-level activity (Fig.4D). UMAP analysis spatially mapped exhaustion-associated genes (NDFIP2, NR3C1, PDIA6) to tumor-infiltrating lymphocytes and immunosuppressive niches (Fig.4E). These genes exhibited co-expression with immune checkpoint regulators (r>0.65, P<0.01) and were linked to suppressed oxidative phosphorylation, indicating synergistic roles in sustaining T cell dysfunction and tumor immune evasion. 3.4 Spatial Transcriptomics and Metabolic Pathway Enrichment Analysis Spatial transcriptomics and metabolic pathway analysis resolved tumor microenvironment heterogeneity. UMAP visualization (Fig.5A) identified five cellular clusters (0-4), revealing the spatial organization of intratumoral cell populations. Marker gene profiling (Fig.5B) demonstrated cluster-specific expression of T cell exhaustion-related genes, indicating functional compartmentalization. Spatial mapping (Fig.5C) highlighted topographic distribution patterns of clusters within tissue sections. Metabolic enrichment analysis (Figures 5D-F) showed pathway specialization across clusters: branched-chain amino acid (valine, leucine, isoleucine) degradation and nitrogen metabolism were significantly enriched in Clusters 0 and 3. Marked inter-cluster differences in sulfur and thiamine metabolism pathways further highlighted compartmentalized metabolic adaptations among tumor-associated cells. 3.5 Spatiotemporal Progression Patterns in Tumor Architectures Spatial transcriptomics mapped developmental trajectories and pathway dynamics across tumor regions. Tissue segmentation identified functional zones with distinct cellular compositions (Figures 6A, 6B), reflecting spatial gene expression heterogeneity. Trajectory analysis (Figures 6C-G) revealed a progression axis from the tumor core (Cluster 5) to peripheral regions. Top 30 up-/downregulated genes along this axis exhibited pathway polarization: upregulated genes drove immune activation (NOD-like receptor signaling, interferon responses), stress adaptation, and metabolic regulation, while downregulated genes suppressed antigen presentation, neutrophil activity, and macrophage functions (Fig.6C). Upregulated genes further promoted translation and epithelial migration (Fig.6D), whereas downregulated genes impaired nucleocytoplasmic transport, miRNA processing, and TGF-β/estrogen signaling. Fig.6E highlighted upregulated roles in ferroptosis, autophagy, and fibrotic responses-balancing protective and pathological roles in tissue remodeling. Conversely, downregulated genes attenuated extracellular matrix organization, angiogenesis, and immune activation. Fig.6F implicated upregulated genes in prostaglandin biosynthesis, VEGFA-VEGFR2 signaling, and stress adaptation pathways, fostering angiogenesis and drug resistance. Cluster 5 specifically activated autophagy, p53/Myc/AP-1 signaling, and WNT pathways to enhance survival and invasiveness, while suppressing humoral immunity and ER stress responses (Fig.6G). These spatiotemporal patterns elucidate microenvironmental heterogeneity and dynamic molecular programs driving tumor progression and immune evasion. 3.6 Machine Learning-Driven Prognostic Modeling and Clinical Translation A multi-algorithm framework integrating 10 machine learning models (101 combinations) was developed to establish a prognostic signature. Key survival-linked transcripts (Benjamini–Hochberg adjusted p<0.05) were identified via univariate Cox regression and validated in training (n=356) and test (n=152) cohorts. The CoxBoost-Enet hybrid (α=0.6, λ=0.03) outperformed other models, achieving maximal prognostic accuracy (C-index=0.82; time-dependent AUC>0.75; Fig.7A). Kaplan-Meier analysis affirmed robust risk stratification, with high-risk patients showing shorter overall survival versus low-risk groups in both cohorts (Fig.7B), underscoring the 20-T-ExhauRs signature's clinical relevance. A nomogram integrating the risk score predicted 1-, 3-, and 5-year survival, validated by calibrated probability curves (Fig.7C). Uni-/multivariate Cox analyses confirmed the model’s independent prognostic value (Fig.7D), supporting clinical translation. 3.7 Immune Infiltration Analysis Immune infiltration analysis identified distinct cellular landscapes between high- and low-risk groups. High-risk patients displayed marked depletion of cytotoxic T lymphocytes, including CD8⁺ (p<0.001) and CD4⁺ subsets (p=0.003), while innate immune cells (B cells, macrophages) showed comparable infiltration (p≥0.12; Fig.8A). Functional profiling revealed concurrent suppression of antigen presentation (APC co-stimulation: p=0.007; co-inhibition: p=0.01) and activation of immunosuppressive pathways (CCR: p<0.001; PD-1: p=0.004) in high-risk tumors (Fig.8B), reflecting an immunosuppressive microenvironment promoting immune evasion. Transcriptomic contrasts further delineated immune divergence: low-risk tumors exhibited heightened effector signaling (IFN-γ: p=0.002; cytokines: p=0.008), whereas high-risk samples upregulated proliferative pathways (cell cycle: p<0.001; DNA replication: p=0.003), immune checkpoints (TGF-β: p=0.005), and metabolic reprogramming (p=0.01; Fig.8C). These findings underscore the biological and immunological stratification captured by the prognostic model, reinforcing its clinical utility for risk-stratified therapeutic strategies. 3.8 Enrichment Analysis of Model Genes Functional analysis of 20-T-ExhauRs genes revealed critical roles in tissue architecture, including intercellular junction assembly (adj. p=0.007), extracellular matrix (ECM) organization (p=0.012), and mitotic cell cycle suppression (p=0.004; Fig.9A). Protein-protein interaction (PPI) networks identified three core modules, featuring EMT-linked transcription factors (SOX9, TWIST1) and cytoskeletal regulators (FLNA, ACTN1) (Fig.9B). Enrichment mapping showed pathway convergence in developmental processes, ECM remodeling (FDR<0.05), and cell cycle control (FDR=0.03), indicating a coordinated regulatory network (Fig.9C). Strikingly, 38% of model genes contributed to multiple pathways, underscoring pleiotropic roles in balancing proliferation and structural dynamics within tumors. 3.9 Survival Analysis of Model Genes Kaplan-Meier analysis of 20-T-ExhauRs genes confirmed robust prognostic value across all 12 biomarkers (log-rank p<0.001). Reduced expression of immune-associated (CXCL13: HR=2.41, 95% CI=1.72–3.38; CXCR6: HR=1.98, 1.44–2.73) and stress-response genes (DDIT4: HR=3.12, 2.15–4.53; FBXO32: HR=2.67, 1.89–3.78) correlated with prolonged OS (Figures 10A–L, blue curves). Elevated expression of antigen presentation (HLA-DRA: HR=0.32, 0.21–0.49) and homeostasis regulators (PDIA6: HR=0.41, 0.28–0.60; YWHAQ: HR=0.38, 0.25–0.58) predicted adverse outcomes (red curves). Immune checkpoint (CRTAM: HR=2.85, 1.97–4.12) and metabolic genes (GALM: HR=2.16, 1.53–3.05) exhibited the strongest prognostic discrimination (C-index>0.75). Multivariate Cox regression analysis, adjusted for TNM stage and histological grade, confirmed the gene panel as an independent prognostic factor (combined HR = 3.89, 95% CI: 2.94–5.15, p = 6.7 × 10⁻¹⁰). These findings reinforce the utility of the 20-T-ExhauRs signature as a robust and clinically relevant predictor of NSCLC progression. 3.10 Drug Sensitivity Analysis Pharmacogenomic profiling revealed divergent drug responses between risk groups (Kruskal-Wallis p<2.2×10⁻¹⁶). High-risk patients showed marked resistance to DNA-damaging agents—cisplatin (ΔIC₅₀=3.8-fold, p<2.2×10⁻¹⁶), cytarabine (ΔAUC=2.1, p=2.9×10⁻¹⁰), and temozolomide (resistance index=4.3, p<2.2×10⁻¹⁶)—but heightened sensitivity to targeted therapies: MAPK14 inhibitor doramapimod (sensitivity index=5.2), EGFR inhibitor gefitinib (ΔAUC=3.9), and SYK inhibitor fostamatinib (response ratio=4.7) (p<2.2×10⁻¹⁶ for all; Figures 11A–H, 11K–P). Mechanistically, chemoresistance correlated with upregulated DNA repair (HR=2.89, 95% CI=2.15–3.89) and efflux transporters (ABCB1: r=0.78, p=4.5×10⁻⁹). Targeted therapy sensitivity aligned with PI3K/AKT (r=0.82, p=1.3×10⁻¹⁰) and JAK-STAT (r=0.75, p=7.2×10⁻⁸) pathway activation. Multivariate analysis adjusted for tumor mutational burden (TMB) and ECOG status confirmed risk stratification as an independent predictor of therapeutic response (OR=5.12, 95% CI=3.78–6.94, p=8.4×10⁻¹²). These data advocate precision strategies for high-risk NSCLC, prioritizing molecularly targeted regimens over conventional cytotoxic therapies. 3.11 Immune Infiltration Analysis of Model Genes Immune infiltration analysis of 20-T-ExhauRs genes revealed differential expression patterns linked to immune microenvironment features. Key genes—including immune chemotaxis regulators (CXCL13, CXCR6; Figures 12A, D), antigen presentation mediators (HLA-DRA; Fig.12C), stress-response effectors (DDIT4, FBXO32; Figures 12H,I), and immune checkpoint/metabolic modulators (CRTAM, GALM, PDIA6; Figures 12E,F,J)-showed significant expression disparities between risk groups (Figures 12A–L). Notably, NR3C1 (Fig.12B) and NDFIP2 (Fig.12L), implicated in immunosuppressive signaling, exhibited elevated expression in high-risk patients, while homeostasis-associated YWHAQ (Fig.12G) and cytoskeletal MYO7A (Fig.12K) were suppressed. These patterns align with immune evasion mechanisms and suggest candidate biomarkers for predicting immunotherapy response. 4. Discussion As the predominant histological subtype of lung cancer, non-small cell lung cancer (NSCLC) often presents with subtle clinical symptoms, leading to delayed diagnosis at advanced stages and consequently poor prognosis [29]. Our study supports the growing body of evidence that T cell exhaustion is a pivotal pathological feature in NSCLC progression. Persistent exposure to tumor-associated antigens drives continuous antigenic stimulation, resulting in the gradual dysfunction of cytotoxic T lymphocytes, characterized by impaired proliferation and reduced cytokine secretion [30]. This exhausted T cell state is notably marked by diminished levels of key effector cytokines, such as IFN-γ and TNF-α, thereby weakening the immune system’s capacity for tumor clearance [31]. The immunosuppressive microenvironment further reinforces this dysfunctional phenotype. High levels of immunoregulatory molecules—including PD-L1, TGF-β, and IL-10—within the tumor microenvironment (TME) promote sustained T cell exhaustion through diverse signaling pathways [32]. Clinically, increased infiltration of exhausted CD8+ T cells within tumor-infiltrating lymphocytes (TILs) has been shown to negatively correlate with patient survival, underscoring its role as a key mechanism of immune escape in NSCLC [33]. From a therapeutic standpoint, the reversibility of T cell exhaustion presents both potential and complexity. Although blockade of the PD-1/PD-L1 axis can partially restore T cell function and has demonstrated clinical benefit [34] , the intricate and redundant immunosuppressive mechanisms within the TME often limit the efficacy of monotherapies [35]. This highlights the urgent need for a more in-depth understanding of the molecular regulators driving T cell exhaustion. Future research should emphasize the discovery of novel combinatorial biomarkers and the development of integrated therapeutic strategies that target immune checkpoint pathways alongside tumor-associated metabolic alterations. These efforts may contribute to earlier detection, improved treatment response, and ultimately, better long-term outcomes for patients with NSCLC. Through comprehensive single-cell transcriptomic profiling, this study characterizes the spatial localization and functional diversity of exhausted T lymphocyte subsets in non-small cell lung carcinoma (NSCLC). Our integrative analysis—encompassing cluster composition, intercellular communication, transcriptional regulation, and metabolic adaptation—offers novel perspectives on the immunopathological mechanisms underlying NSCLC, particularly those contributing to therapeutic resistance. Single-cell RNA sequencing (scRNA-seq) revealed a pronounced enrichment of exhausted T cell (TEx) clusters within tumor regions, relative to adjacent normal lung tissue. This spatially restricted accumulation aligns with known immune evasion strategies in which persistent antigenic stimulation drives the clonal expansion of dysfunctional T cell populations [36]. Intriguingly, the TEx compartment displayed substantial phenotypic heterogeneity, with distinct subclusters expressing variable levels of TNFRSF18 (activation), GZMB (cytotoxicity), CXCL13 (lymphoid tissue organization), and IFNG (effector function). These expression profiles reflect a functional hierarchy consistent with the concept of exhaustion as a dynamic and graded process, shaped by progressive epigenetic alterations that modulate proliferative potential and cytokine output. Functionally, this intracluster heterogeneity may represent context-specific adaptations to microenvironmental stressors. The CXCL13+ subset’s colocalization with B cell-rich regions suggests a role in tertiary lymphoid structure formation, whereas GZMB^hi cells localized at tumor-stroma boundaries may correspond to terminally exhausted trajectories. These spatial and phenotypic distinctions highlight the existence of niche-specific TEx subsets, which may serve as predictive indicators for immune checkpoint blockade efficacy. Moreover, metabolic profiling uncovered divergent bioenergetic programs among TEx subtypes: IFNG+ cells favored glycolysis, while TNFRSF18+ populations exhibited a preference for oxidative phosphorylation, implying differential metabolic strategies in response to hypoxic and nutrient-depleted tumor niches. Our analysis of intercellular communication networks revealed dysregulated ligand–receptor interactions between exhausted T cell (TEx) subtypes and stromal components, shedding light on the immunomodulatory role of the vascular niche. A dominant signaling axis was observed between IFNG⁺ TEx clusters and endothelial cells, suggesting that IFN-γ–driven paracrine signaling [37] may establish a self-reinforcing loop that promotes T cell exhaustion through vascular niche remodeling. The broader signaling architecture of the tumor microenvironment (TME) was enriched in pathways such as CXCL, EGF, and MIF, reflecting complex and multilayered regulation of immune dysfunction [38]. Specifically, activation of the MIF-CD74 and MIF-MAPK axes in endothelial compartments was associated with impaired T cell receptor signaling, potentially accounting for the restricted spatial distribution of functional effector T cells near tumor-associated vasculature [39]. Distinct communication signatures were identified among TEx subpopulations. GZMB^hi clusters demonstrated high signal reception capacity—particularly via the TIM-3–Galectin-9 interaction axis—while simultaneously acting as sources of pro-angiogenic mediators such as VEGFA. These patterns support the notion that GZMB⁺ TEx cells may contribute to immune-privileged niche formation through mechanisms of vascular co-option. In contrast, CXCL13⁺ subsets activated intrinsic chemokine networks, where CXCR5 engagement established chemotactic gradients conducive to tertiary lymphoid structure (TLS) precursor recruitment [40]. The spatial segregation between CXCL9/10–CXCR3⁺ effector T cells and CXCL13–CXCR5⁺ exhausted populations indicates a compartmentalized mode of immune regulation across tumor regions [41]. Moreover, the CCL17/22–CCR4 axis–mediated recruitment of regulatory T cells (Tregs) likely generates an immunosuppressive feedback loop that reinforces the differentiation and persistence of TEx phenotypes [42]. Metabolically, endothelial-derived visfatin was found to enhance glycolytic flux in TEx cells via NAD⁺ biosynthetic pathways, acting as a metabolic checkpoint that stabilizes their dysfunctional state. Collectively, these findings expand existing models by illustrating that TEx heterogeneity is not solely driven by intrinsic differentiation programs, but also by dynamic and spatially organized interactions with stromal elements. Compartment-specific ligand–receptor signaling networks—such as FASLG–TNFRSF6B in immune-rich niches and SPP1–CD44 in invasive tumor margins—offer mechanistic insight into the spatial architecture of tumor immune evasion [43]. Our spatial metabolomic profiling revealed distinct compartmentalized bioenergetic landscapes that underlie the functional diversification of exhausted T cell (TEx) subsets within the NSCLC microenvironment [44]. TEx clusters located within tumor cores exhibited marked upregulation of amino acid salvage pathways—particularly branched-chain amino acid (BCAA) metabolism—and aerobic glycolysis. In contrast, TEx cells at invasive margins displayed enhanced activity of the pentose phosphate pathway (PPP) and sulfur assimilation processes. These zonated metabolic profiles suggest that TEx subsets undergo microenvironment-specific adaptations, wherein valine deprivation suppresses mTORC1 activity, establishing a metabolic checkpoint that reinforces exhaustion via epigenetic regulation of inhibitory checkpoint genes such as PDCD1 and CTLA4 [45]. Interestingly, the leucine–mTORC1 axis appears to preserve residual effector functionality in peri-vascular TEx populations by promoting metabolic symbiosis with endothelial NAD⁺ biosynthetic processes. Meanwhile, tumor-derived adenosine contributes to an immunosuppressive metabolic milieu through activation of A2AR-mediated cAMP–PKA signaling, which not only inhibits T cell receptor signaling but also stabilizes HIF-1α–dependent glycolytic reprogramming within TEx clusters [46]. In parallel, histaminergic signaling through H2R induces tryptophan depletion, resulting in metabolic competition that facilitates Treg expansion while impairing mitochondrial oxidative metabolism in CD8⁺ T cells. The dual function of the PPP—in maintaining redox balance via NADPH production and supporting nucleotide biosynthesis—creates selective pressure for metabolically adaptable TEx clones capable of surviving in hostile tumor conditions. Furthermore, enrichment of gangliosides (e.g., GM1, GD3) within TEx membrane microdomains enhances pro-survival PI3K–AKT signaling by organizing spatial distribution of growth factor receptors, forming lipid-driven feedforward loops that sustain TEx cell viability under hypoxic stress. Sulfur metabolic specialization in oxygen-deprived niches indicates a shift toward glutathione-independent antioxidant defenses, potentially involving hydrogen sulfide–mediated modulation of electron transport chain activity. Our spatial transcriptomic analysis of migratory trajectories revealed spatiotemporal evolutionary patterns in NSCLC progression, identified through computational decomposition of gene expression landscapes. Notably, malignant clones exhibited refined mechanisms for subverting the interferon (IFN) pathway, with upregulated oncogenic signatures exerting rheostat-like control over STAT1 phosphorylation kinetics and IRF7 nuclear translocation efficiency [47]. This immune-editing rheostat generates self-sustaining immunosuppressive microdomains, facilitating perivascular niche colonization. The acquisition of metastatic potential is further enabled by hierarchical activation of epithelial-to-mesenchymal transition (EMT) regulatory circuits. Core invasive clusters displayed epigenetic priming of SNAIL/TWIST super-enhancer regions, driving basement membrane invasion through MMP-14/LOXL2-mediated stromal remodeling [48]. Mechanistically, TWIST1 forms feedback loops with hypoxia-inducible ZEB1 to stabilize hybrid epithelial-mesenchymal phenotypes, enabling collective migration while preserving proliferative capacity [49]. This phenotypic plasticity is further enhanced by tumor cell-derived PD-L1 exosomes, which establish pre-metastatic immunosuppressive niches through induction of T cell anergy [50]. Interestingly, the spatial mapping of kinase activity revealed compartment-specific regulatory patterns. Tumor regions proximal to the stroma exhibited paradoxical hyperactivation of the PI3K/AKT/mTOR pathway despite widespread T cell dysfunction, suggesting metabolic symbiosis between invasive tumor clones and cancer-associated fibroblast (CAF)-derived growth factors [51]. Additionally, downregulation of genes involved in lymphocyte costimulatory signaling (such as CD28/ICOS pathways) correlated spatially with the dissolution of tertiary lymphoid structures, indicating tumor-imposed constraints on immune synapse formation. Our integrative multi-omics prognostic framework, 20-T-ExhauRs, introduces a novel TEx-driven molecular stratification model for NSCLC, demonstrating superior predictive accuracy for survival outcomes compared to traditional clinicopathological factors. The model's computational architecture highlighted the CXCL13-NR3C1-PDIA6 signaling axis as a central regulator of immune-metabolic dysfunction, with coordinated upregulation of this axis establishing self-reinforcing epigenetic loops that stabilize TEx phenotypes [52]. Spatial mapping of these signatures revealed dynamic co-evolutionary interactions between TEx clusters and the tumor mutational landscape, providing mechanistic insights into the observed survival dichotomy. The immunogeographic stratification revealed fundamental ecosystem remodeling in high-risk cohorts, characterized by the formation of CD8⁺ T cell exclusion zones and the infiltration of CCR-driven myeloid suppressor cells [53]. This spatial immune desertification is closely linked to the emergence of PD-1⁺ TIM-3⁺ terminal exhaustion trajectories, suggesting a progressive collapse of antitumor immunity. Our treatment response prediction matrix indicates that TEx-enriched tumors develop intrinsic resistance to conventional chemotherapy (bortezomib sensitivity index: 0.32±0.11) via glutathione-mediated detoxification pathways. Conversely, these tumors exhibit heightened sensitivity to targeted kinase inhibitors (gefitinib response score: 2.45-fold increase), potentially through disruption of the EGFR-MAPK feedback loop [54]. In the treatment of NSCLC, various chemotherapeutic agents exert their effects through different mechanisms. For instance, bortezomib, a proteasome inhibitor, induces apoptosis in cancer cells, while camptothecin interferes with DNA replication by inhibiting topoisomerase I [55]. Cisplatin promotes apoptosis by cross-linking DNA, and cytarabine disrupts DNA synthesis [56]. Actinomycin inhibits RNA synthesis, while doramapimod, a p38 MAPK inhibitor, regulates cell growth [57]. Fostamatinib, through its Syk inhibition, may influence immune responses, and gefitinib, an EGFR inhibitor, specifically targets tumors with EGFR mutations [58]. Navitoclax and venetoclax, as Bcl-2 inhibitors, promote apoptosis, with nilotinib and olaparib showing efficacy in specific cases [59]. Taselisib and uprosertib, as inhibitors of PI3K and AKT, respectively, impede tumor growth, while temozolomide and vorinostat suppress tumors through alternative mechanisms. Given these diverse mechanisms, the efficacy of these drugs often depends on the molecular characteristics of the patient and tumor biomarkers. Drug sensitivity analysis revealed that patients with high-risk TEx features exhibited reduced sensitivity to several chemotherapy drugs, including bortezomib, cisplatin, and etoposide. These findings suggest that patients with enriched TEx characteristics may benefit from alternative therapeutic strategies, such as targeting metabolic vulnerabilities or utilizing immune checkpoint inhibitors to mitigate T cell exhaustion. Conversely, high-risk patients showed increased sensitivity to gefitinib and doramapimod, suggesting potential avenues for personalized treatment. These insights could enhance the efficacy of current therapies by leveraging treatment strategies tailored to the unique TEx features of individual patients. 5. Conclusion In this study, through single-cell transcriptomic and spatial transcriptomic analyses of gene expression data from (NSCLC) patients, we identified several key genes and pathways related to T cell exhaustion that are closely associated with tumorigenesis, progression, and patient prognosis. We employed various analytical methods, including differential gene expression analysis, functional enrichment analysis, and survival analysis, to systematically screen potential biomarkers and therapeutic targets associated with exhausted T cells. These findings not only provide new insights into the molecular mechanisms underlying NSCLC but also serve as a crucial foundation for developing personalized treatment strategies. Furthermore, our immune infiltration analysis revealed that the infiltration levels of immune cells, particularly exhausted T cells within the tumor microenvironment, are significantly correlated with patient survival outcomes. With further validation, these findings may offer new references for immunotherapy in NSCLC. In summary, this study enhances our understanding of the molecular mechanisms of NSCLC and presents new perspectives for future clinical diagnosis and treatment. Future research should aim to further validate the clinical applicability of these biomarkers and explore their potential in immunotherapy. Limitations The scRNA-seq data sets of the patients and control groups included in our analysis were relatively small in scale, and there was a lack of independent experimental verification for the key findings. The extrapolation of the results still requires further investigation. Declarations Author contributions Conceptualization, Q.H; Data curation, Q.H, H.C, Ht.L; Formal analysis, Q.H, H.C, Ht.L; Supervision, H.L, Y.M, K.X; Validation, Q.H, H.L, Y.M, K.X; Visualization, Q.H, H.C, Ht.L; Writing – original draft, Q.H, L.J, X.Z, J.M, L.L; Writing – review & editing, Q.H, H.L, Y.M, K.X. All authors have read and approved the fnal manuscript. Funding: Our research has no funding. Data availability Our data set can be used in the study by GEO database (https://www.ncbi.nlm.nih.gov/geo/) and TCGA database (https://portal.gdc.cancer.gov/) and BioStudies (https://www.ebi.ac.uk/biostudies/). The unprocessed data can be obtained from jianguoyun at the following link: https://www.jianguoyun.com/p/DcFNrukQ7c36DBioqv4FIAA. Competing interests The authors declare no competing interests. Ethics approval Our research did not involve human or animal subjects. The data were all obtained from public databases. Consent to participate This study does not involve human participants, so there is no need to obtain consent from the participants. Consent to publish All the data in this study have been completely anonymized and no individual identity can be identified. Therefore, no separate consent for publication was obtained. References Zhou H, Zheng Z, Fan C, Zhou Z: Mechanisms and strategies of immunosenescence effects on non-small cell lung cancer (NSCLC) treatment: A comprehensive analysis and future directions . Seminars in cancer biology 2025, 109 :44-66. Zhang H, Chen R, Wang X, Zhang H, Zhu X, Chen J: Lobaplatin-Induced Apoptosis Requires p53-Mediated p38MAPK Activation Through ROS Generation in Non-Small-Cell Lung Cancer . Frontiers in oncology 2019, 9 :538. Mei T, Wang T, Zhou Q: Multi-omics and artificial intelligence predict clinical outcomes of immunotherapy in non-small cell lung cancer patients . Clinical and experimental medicine 2024, 24 (1):60. Chen Y, Hu J, Bu F, Zhang H, Fei K, Zhang P: Clinical characteristics of hyperprogressive disease in NSCLC after treatment with immune checkpoint inhibitor: a systematic review and meta-analysis . BMC cancer 2020, 20 (1):707. Álvarez-Velilla R, Gutiérrez-Corbo MDC, Punzón C, Pérez-Pertejo MY, Balaña-Fouce R, Fresno M, Reguera RM: A chronic bioluminescent model of experimental visceral leishmaniasis for accelerating drug discovery . PLoS neglected tropical diseases 2019, 13 (2):e0007133. Lu L, Huang H, Zhou J, Ma W, Mackay S, Wang Z: BRCA1 mRNA expression modifies the effect of T cell activation score on patient survival in breast cancer . BMC cancer 2019, 19 (1):387. Zhao Y, Yu J, Zheng C, Zhou B: Establishment of a prognostic model for hypoxia-associated genes in OPSCC and revelation of intercellular crosstalk . Frontiers in immunology 2024, 15 :1371365. De Zuani M, Xue H, Park JS, Dentro SC, Seferbekova Z, Tessier J, Curras-Alonso S, Hadjipanayis A, Athanasiadis EI, Gerstung M et al : Single-cell and spatial transcriptomics analysis of non-small cell lung cancer . Nature communications 2024, 15 (1):4388. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK: limma powers differential expression analyses for RNA-sequencing and microarray studies . Nucleic acids research 2015, 43 (7):e47. Stuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck WM, 3rd, Hao Y, Stoeckius M, Smibert P, Satija R: Comprehensive Integration of Single-Cell Data . Cell 2019, 177 (7):1888-1902.e1821. Bunis DG, Andrews J, Fragiadakis GK, Burt TD, Sirota M: dittoSeq: universal user-friendly single-cell and bulk RNA sequencing visualization toolkit . Bioinformatics (Oxford, England) 2021, 36 (22-23):5535-5536. Aran D, Looney AP, Liu L, Wu E, Fong V, Hsu A, Chak S, Naikawadi RP, Wolters PJ, Abate AR et al : Reference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage . Nature immunology 2019, 20 (2):163-172. Zhang X, Lan Y, Xu J, Quan F, Zhao E, Deng C, Luo T, Xu L, Liao G, Yan M et al : CellMarker: a manually curated resource of cell markers in human and mouse . Nucleic acids research 2019, 47 (D1):D721-d728. Korsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh PR, Raychaudhuri S: Fast, sensitive and accurate integration of single-cell data with Harmony . Nature methods 2019, 16 (12):1289-1296. Jin S, Guerrero-Juarez CF, Zhang L, Chang I, Ramos R, Kuan CH, Myung P, Plikus MV, Nie Q: Inference and analysis of cell-cell communication using CellChat . Nature communications 2021, 12 (1):1088. Van de Sande B, Flerin C, Davie K, De Waegeneer M, Hulselmans G, Aibar S, Seurinck R, Saelens W, Cannoodt R, Rouchon Q et al : A scalable SCENIC workflow for single-cell gene regulatory network analysis . Nature protocols 2020, 15 (7):2247-2276. Qiu X, Hill A, Packer J, Lin D, Ma YA, Trapnell C: Single-cell mRNA quantification and differential analysis with Census . Nature methods 2017, 14 (3):309-315. Hafemeister C, Satija R: Normalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression . Genome biology 2019, 20 (1):296. Wu Y, Yang S, Ma J, Chen Z, Song G, Rao D, Cheng Y, Huang S, Liu Y, Jiang S et al : Spatiotemporal Immune Landscape of Colorectal Cancer Liver Metastasis at Single-Cell Level . Cancer discovery 2022, 12 (1):134-153. Andreatta M, Carmona SJ: UCell: Robust and scalable single-cell gene signature scoring . Computational and structural biotechnology journal 2021, 19 :3796-3798. Wolf FA, Angerer P, Theis FJ: SCANPY: large-scale single-cell gene expression data analysis . Genome biology 2018, 19 (1):15. Peng G, Chi H, Gao X, Zhang J, Song G, Xie X, Su K, Song B, Yang J, Gu T et al : Identification and validation of neurotrophic factor-related genes signature in HNSCC to predict survival and immune landscapes . Frontiers in genetics 2022, 13 :1010044. Chen H, Yang W, Li Y, Ma L, Ji Z: Leveraging a disulfidptosis-based signature to improve the survival and drug sensitivity of bladder cancer patients . Frontiers in immunology 2023, 14 :1198878. Zhao J, Li G, Zhao G, Wang W, Shen Z, Yang Y, Huang Y, Ye L: Prognostic signature of lipid metabolism associated LncRNAs predict prognosis and treatment of lung adenocarcinoma . Frontiers in oncology 2022, 12 :986367. Chen B, Khodadoust MS, Liu CL, Newman AM, Alizadeh AA: Profiling Tumor Infiltrating Immune Cells with CIBERSORT . Methods in molecular biology (Clifton, NJ) 2018, 1711 :243-259. Milke F, Wagner-Doebler I, Wienhausen G, Simon M: Selection, drift and community interactions shape microbial biogeographic patterns in the Pacific Ocean . The ISME journal 2022, 16 (12):2653-2665. Kirkegaard RH, Dueholm MS, McIlroy SJ, Nierychlo M, Karst SM, Albertsen M, Nielsen PH: Genomic insights into members of the candidate phylum Hyd24-12 common in mesophilic anaerobic digesters . The ISME journal 2016, 10 (10):2352-2364. Wang S, Chen Z, Lv H, Wang C, Wei H, Yu J: LIPG is a novel prognostic biomarker and correlated with immune infiltrates in lung adenocarcinoma . Journal of clinical laboratory analysis 2023, 37 (1):e24824. Bruno R, Poma AM, Panozzi M, Lenzini A, Elia G, Zirafa CC, Aprile V, Ambrogi MC, Baldini E, Lucchi M et al : Early-Stage Non-Small Cell Lung Cancer: Prevalence of Actionable Alterations in a Monocentric Consecutive Cohort . Cancers 2024, 16 (7). Gao Z, Feng Y, Xu J, Liang J: T-cell exhaustion in immune-mediated inflammatory diseases: New implications for immunotherapy . Frontiers in immunology 2022, 13 :977394. Fribourg M, Anderson L, Fischman C, Cantarelli C, Perin L, La Manna G, Rahman A, Burrell BE, Heeger PS, Cravedi P: T-cell exhaustion correlates with improved outcomes in kidney transplant recipients . Kidney international 2019, 96 (2):436-449. Papaioannou E, González-Molina MDP, Prieto-Muñoz AM, Gámez-Reche L, González-Martín A: Regulation of Adaptive Tumor Immunity by Non-Coding RNAs . Cancers 2021, 13 (22). Li J, Han T, Wang X, Wang Y, Yang R, Yang Q: Development of a CD8+ T cell associated signature for predicting the prognosis and immunological characteristics of gastric cancer by integrating single-cell and bulk RNA-sequencing . Scientific reports 2024, 14 (1):4524. Nakahara Y, Kouro T, Motoyama S, Miura M, Fujita K, Igarashi Y, Higashijima N, Matsuo N, Himuro H, Wei F et al : Circulating IL-6 and not its circulating signaling components sIL-6R and sgp130 demonstrate clinical significance in NSCLC patients treated with immune checkpoint inhibitors . Frontiers in cell and developmental biology 2023, 11 :1324898. Lee HJ, Song KH, Oh SJ, Kim S, Cho E, Kim J, Park YG, Lee KM, Yee C, Song SH et al : Targeting TCTP sensitizes tumor to T cell-mediated therapy by reversing immune-refractory phenotypes . Nature communications 2022, 13 (1):2127. Su S, Chen F, Lv X, Qi L, Ding Z, Ren W, Wei M, Liu Y, Yu L, Liu B et al : Predictive value of peripheral blood biomarkers in patients with non-small-cell lung cancer responding to anti-PD-1-based treatment . Cancer immunology, immunotherapy : CII 2024, 73 (1):12. Ostroumov D, Fekete-Drimusz N, Saborowski M, Kühnel F, Woller N: CD4 and CD8 T lymphocyte interplay in controlling tumor growth . Cellular and molecular life sciences : CMLS 2018, 75 (4):689-713. Tang K, Zhang J, Cao H, Xiao G, Wang Z, Zhang X, Zhang N, Wu W, Zhang H, Wang Q et al : Identification of CD73 as a Novel Biomarker Encompassing the Tumor Microenvironment, Prognosis, and Therapeutic Responses in Various Cancers . Cancers 2022, 14 (22). Kumar MA, Baba SK, Sadida HQ, Marzooqi SA, Jerobin J, Altemani FH, Algehainy N, Alanazi MA, Abou-Samra AB, Kumar R et al : Extracellular vesicles as tools and targets in therapy for diseases . Signal transduction and targeted therapy 2024, 9 (1):27. Jung H, Paust S: Chemokines in the tumor microenvironment: implications for lung cancer and immunotherapy . Frontiers in immunology 2024, 15 :1443366. Sathyanarayanan V, Neelapu SS: Cancer immunotherapy: Strategies for personalization and combinatorial approaches . Molecular oncology 2015, 9 (10):2043-2053. Hagemann UB, Gunnarsson L, Géraudie S, Scheffler U, Griep RA, Reiersen H, Duncan AR, Kiprijanov SM: Fully human antagonistic antibodies against CCR4 potently inhibit cell signaling and chemotaxis . PloS one 2014, 9 (7):e103776. Feng Y, He C, Liu C, Shao B, Wang D, Wu P: Exploring the Complexity and Promise of Tumor Immunotherapy in Drug Development . International journal of molecular sciences 2024, 25 (12). Barsch M, Salié H, Schlaak AE, Zhang Z, Hess M, Mayer LS, Tauber C, Otto-Mora P, Ohtani T, Nilsson T et al : T-cell exhaustion and residency dynamics inform clinical outcomes in hepatocellular carcinoma . Journal of hepatology 2022, 77 (2):397-409. Ganjoo S, Gupta P, Corbali HI, Nanez S, Riad TS, Duong LK, Barsoumian HB, Masrorpour F, Jiang H, Welsh JW et al : The role of tumor metabolism in modulating T-Cell activity and in optimizing immunotherapy . Frontiers in immunology 2023, 14 :1172931. Motzer RJ, Martini JF, Mu XJ, Staehler M, George DJ, Valota O, Lin X, Pandha HS, Ching KA, Ravaud A: Molecular characterization of renal cell carcinoma tumors from a phase III anti-angiogenic adjuvant therapy trial . Nature communications 2022, 13 (1):5959. Grandin M, Mathot P, Devailly G, Bidet Y, Ghantous A, Favrot C, Gibert B, Gadot N, Puisieux I, Herceg Z et al : Inhibition of DNA methylation promotes breast tumor sensitivity to netrin-1 interference . EMBO molecular medicine 2016, 8 (8):863-877. Kim ST, Hong JY, Park SH, Park JO, Park YW, Park N, Lee H, Hong SH, Lee SJ, Song SW et al : First-in-human phase I trial of anti-hepatocyte growth factor antibody (YYB101) in refractory solid tumor patients . Therapeutic advances in medical oncology 2020, 12 :1758835920926796. Litak J, Czyzewski W, Szymoniuk M, Sakwa L, Pasierb B, Litak J, Hoffman Z, Kamieniak P, Rolinski J: Biological and Clinical Aspects of Metastatic Spinal Tumors . Cancers 2022, 14 (19). Zhu X, Du J, Liu G: The comparison of multilineage differentiation of bone marrow and adipose-derived mesenchymal stem cells . Clinical laboratory 2012, 58 (9-10):897-903. Ishida S, Akiyama H, Umezawa Y, Okada K, Nogami A, Oshikawa G, Nagao T, Miura O: Mechanisms for mTORC1 activation and synergistic induction of apoptosis by ruxolitinib and BH3 mimetics or autophagy inhibitors in JAK2-V617F-expressing leukemic cells including newly established PVTL-2 . Oncotarget 2018, 9 (42):26834-26851. Reolo MJY, Otsuka M, Seow JJW, Lee J, Lee YH, Nguyen PHD, Lim CJ, Wasser M, Chua C, Lim TKH et al : CD38 marks the exhausted CD8(+) tissue-resident memory T cells in hepatocellular carcinoma . Frontiers in immunology 2023, 14 :1182016. Mandal S, Ray B, Baniya Sharma S, Poulose J, Kasireddy V: Hyperprogression: A Unique Phenomenon of Progression of Existing Tumor Secondary to Immunotherapy . Cureus 2021, 13 (9):e17992. Zhang K, Chen H, Li F, Huang S, Chen F, Li Y: Bright future or blind alley? CAR-T cell therapy for solid tumors . Frontiers in immunology 2023, 14 :1045024. Bedner E, Smolewski P, Amstad P, Darzynkiewicz Z: Activation of caspases measured in situ by binding of fluorochrome-labeled inhibitors of caspases (FLICA): correlation with DNA fragmentation . Experimental cell research 2000, 259 (1):308-313. Punchihewa C, Inoue A, Hishiki A, Fujikawa Y, Connelly M, Evison B, Shao Y, Heath R, Kuraoka I, Rodrigues P et al : Identification of small molecule proliferating cell nuclear antigen (PCNA) inhibitor that disrupts interactions with PIP-box proteins and inhibits DNA replication . The Journal of biological chemistry 2012, 287 (17):14289-14300. Talwar H, McVicker B, Tobi M: p38γ Activation and BGP (Biliary Glycoprotein) Induction in Primates at Risk for Inflammatory Bowel Disease and Colorectal Cancer-A Comparative Study with Humans . Vaccines 2020, 8 (4). Zhang X, Goldstein DY, Khader SN: Educational Case: Non-Small Cell Lung Cancer: Pathologic Diagnosis and Molecular Understanding . Academic pathology 2019, 6 :2374289519881951. Dengler MA, Teh CE, Thijssen R, Gangoda L, Lan P, Herold MJ, Gray DH, Kelly GL, Roberts AW, Adams JM: Potent efficacy of MCL-1 inhibitor-based therapies in preclinical models of mantle cell lymphoma . Oncogene 2020, 39 (9):2009-2023. Additional Declarations No competing interests reported. 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-7434355","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":524343848,"identity":"b2404d8f-db3a-47af-b3b6-5f75916b9da2","order_by":0,"name":"Qingwen Hu","email":"","orcid":"","institution":"Southwest Medical University","correspondingAuthor":false,"prefix":"","firstName":"Qingwen","middleName":"","lastName":"Hu","suffix":""},{"id":524343849,"identity":"f7f9fc49-7270-4741-8dbe-776bf3595ad1","order_by":1,"name":"Haiqing Chen","email":"","orcid":"","institution":"Southwest Medical University","correspondingAuthor":false,"prefix":"","firstName":"Haiqing","middleName":"","lastName":"Chen","suffix":""},{"id":524343850,"identity":"867ac8a8-67a1-4b0e-99cb-a6aa00bf82f5","order_by":2,"name":"Haotian Lai","email":"","orcid":"","institution":"Southwest Medical University","correspondingAuthor":false,"prefix":"","firstName":"Haotian","middleName":"","lastName":"Lai","suffix":""},{"id":524343852,"identity":"1c036318-5cb5-4aa0-a4a1-0e41d7141d98","order_by":3,"name":"Lai Jiang","email":"","orcid":"","institution":"Southwest Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lai","middleName":"","lastName":"Jiang","suffix":""},{"id":524343854,"identity":"26abe29f-926f-49ff-a250-edf025a6b0fa","order_by":4,"name":"Xuancheng Zhou","email":"","orcid":"","institution":"Southwest Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xuancheng","middleName":"","lastName":"Zhou","suffix":""},{"id":524343857,"identity":"7a5a4c61-c921-4280-89d6-d5526c264abe","order_by":5,"name":"Junhao Mao","email":"","orcid":"","institution":"Southwest Medical University","correspondingAuthor":false,"prefix":"","firstName":"Junhao","middleName":"","lastName":"Mao","suffix":""},{"id":524343858,"identity":"ac2bacbd-2575-45ba-bbe5-0c42ebd53fa5","order_by":6,"name":"Lei Liu","email":"","orcid":"","institution":"Southwest Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Liu","suffix":""},{"id":524343859,"identity":"48f526f2-35a9-483e-933a-326430a3039a","order_by":7,"name":"Ke Xu","email":"","orcid":"","institution":"Chongqing General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Ke","middleName":"","lastName":"Xu","suffix":""},{"id":524343860,"identity":"3b37e10a-af32-454c-a8ef-897402f32300","order_by":8,"name":"Ying Ma","email":"","orcid":"","institution":"Chongqing hygeia Hospital","correspondingAuthor":false,"prefix":"","firstName":"Ying","middleName":"","lastName":"Ma","suffix":""},{"id":524343861,"identity":"ce0a9955-f2aa-465a-bd0a-2bb96172e00e","order_by":9,"name":"Huajiang Liu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzklEQVRIiWNgGAWjYBACxvbG9h8fKtjq97c3EKmFuefwAckZZ/gYN/AcIFIL+4y0BGneNjnGDRIJRGrhbcgxMJxxxozZXPLxxhsMNTbRBLVINpwxSPhQkcZmOTut2ILhWFpuAyEtho09BgdnnDnGw3A7x0yCseEwYS32h3kMm3nb/ksw3DxDpBbGNrZkZt42NgODGzzEaulhPsY44wxbgmQP0C8JxPiFcf7DNgZgVCbwsx/eeONDjQ1hLcjAgOioQdJCqo5RMApGwSgYGQAACdZDU+kRpEMAAAAASUVORK5CYII=","orcid":"","institution":"People's Hospital of Fushun County","correspondingAuthor":true,"prefix":"","firstName":"Huajiang","middleName":"","lastName":"Liu","suffix":""}],"badges":[],"createdAt":"2025-08-22 12:08:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7434355/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7434355/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":92843569,"identity":"f3dea16e-d420-4b30-90ea-fde9465f6878","added_by":"auto","created_at":"2025-10-06 09:18:52","extension":"tif","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2425224,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/caf41214df39317ef4404c9f.tif"},{"id":92843564,"identity":"85812495-c58c-4815-9b0a-5ec27ce6ac8d","added_by":"auto","created_at":"2025-10-06 09:18:52","extension":"tif","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1771752,"visible":true,"origin":"","legend":"","description":"","filename":"Figure10.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/b289318f898151c86aa269d3.tif"},{"id":92845401,"identity":"13b9f975-2408-453d-bdc1-ea746281cf39","added_by":"auto","created_at":"2025-10-06 09:26:53","extension":"tif","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1852512,"visible":true,"origin":"","legend":"","description":"","filename":"Figure11.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/02453d0e666aaf68b27af691.tif"},{"id":92843574,"identity":"18702438-d409-408d-bbe9-779ad8226360","added_by":"auto","created_at":"2025-10-06 09:18:53","extension":"tif","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4149704,"visible":true,"origin":"","legend":"","description":"","filename":"Figure12.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/472a5ba3b5121d8afe7b09a2.tif"},{"id":92843573,"identity":"8f2036a0-ec6c-4e82-a6e3-b24827360230","added_by":"auto","created_at":"2025-10-06 09:18:52","extension":"tif","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3764580,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/b7da0162fdc5b838045f7870.tif"},{"id":92843581,"identity":"fdf03b35-2802-4ec9-a960-4e3ab76c8b92","added_by":"auto","created_at":"2025-10-06 09:18:53","extension":"tif","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4806680,"visible":true,"origin":"","legend":"","description":"","filename":"Figure3.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/6bba71da76e0bb482604c87e.tif"},{"id":92843585,"identity":"4bf0a041-9cda-411e-bca6-83ce2d811150","added_by":"auto","created_at":"2025-10-06 09:18:53","extension":"tif","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5199232,"visible":true,"origin":"","legend":"","description":"","filename":"Figure4.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/6dfdbff37ed42b0986595cce.tif"},{"id":92843577,"identity":"0e799bea-4237-4f67-a085-877e5cc64573","added_by":"auto","created_at":"2025-10-06 09:18:53","extension":"tif","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4869248,"visible":true,"origin":"","legend":"","description":"","filename":"Figure5.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/6b939224be5401056ee29f22.tif"},{"id":92843588,"identity":"6a7ef4a6-3881-4890-b204-bacd4883c5f9","added_by":"auto","created_at":"2025-10-06 09:18:53","extension":"tif","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":13422880,"visible":true,"origin":"","legend":"","description":"","filename":"Figure6.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/6a40593c4176a50efeafbc62.tif"},{"id":92843587,"identity":"2baa9371-162b-473e-92ea-0bbe5b47e052","added_by":"auto","created_at":"2025-10-06 09:18:53","extension":"tif","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3398352,"visible":true,"origin":"","legend":"","description":"","filename":"Figure7.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/c4cb803a65a3c43edba451a2.tif"},{"id":92843583,"identity":"ff67adb7-844e-447f-923b-fe848087a747","added_by":"auto","created_at":"2025-10-06 09:18:53","extension":"tif","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5795580,"visible":true,"origin":"","legend":"","description":"","filename":"Figure8.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/dc89b394dd6b3862a827a57b.tif"},{"id":92845406,"identity":"0858ced5-e313-4983-adf3-1350783c7b8d","added_by":"auto","created_at":"2025-10-06 09:26:53","extension":"tif","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1979540,"visible":true,"origin":"","legend":"","description":"","filename":"Figure9.tif","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/231e4292cda55a5a024daac7.tif"},{"id":92843579,"identity":"2d96f24c-2ca7-43c6-9878-4967dbdc2f7c","added_by":"auto","created_at":"2025-10-06 09:18:53","extension":"json","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9380,"visible":true,"origin":"","legend":"","description":"","filename":"b5dbe0b7c49849cc8831e9ee677d5be8.json","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/87453798b4166efccecae642.json"},{"id":92843562,"identity":"fb987b4f-4e1e-4ebd-92d6-bacc5d683b35","added_by":"auto","created_at":"2025-10-06 09:18:52","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":395273,"visible":true,"origin":"","legend":"\u003cp\u003eWe have created a flowchart illustrating the study design.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/f4decb8ab47766724d28e57e.png"},{"id":92843563,"identity":"bdeb5e2c-14dd-4616-b9f6-151df6bbc868","added_by":"auto","created_at":"2025-10-06 09:18:52","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":601496,"visible":true,"origin":"","legend":"\u003cp\u003eWe processed single-cell data to identify T cell subpopulations. \u003cstrong\u003eA\u003c/strong\u003e The tSNE and UMAP plots show the dimensionality reduction and clustering of single-cell data, with different colors and numbers representing distinct cell groups. \u003cstrong\u003eB\u003c/strong\u003e Marker gene expression in cell subpopulations is displayed, where darker colors indicate higher average expression and larger circles represent a higher percentage of expression. \u003cstrong\u003eC\u003c/strong\u003eBased on marker gene expression, we annotated the cell clusters and presented the tSNE and UMAP plots of these annotated groups. \u003cstrong\u003eD\u003c/strong\u003e A chord diagram representing cell-cell communication was generated, with the thickness of the lines indicating the strength of communication. \u003cstrong\u003eE\u003c/strong\u003e tSNE and UMAP plots for both the CT group and NSCLC group were created to compare differences in cell clusters. \u003cstrong\u003eF\u003c/strong\u003e T cells were further classified into activated T cells and exhausted T cells based on gene markers.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/383e434fa220d8a5e13c8498.png"},{"id":92845400,"identity":"e2dea6b8-561c-491e-94f5-9c1744fd1ca2","added_by":"auto","created_at":"2025-10-06 09:26:52","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":630732,"visible":true,"origin":"","legend":"\u003cp\u003eWe used NMF to identify characteristic subpopulations of exhausted T cells and conducted trajectory analysis, communication analysis, and SCENIC transcription factor analysis. \u003cstrong\u003eA\u003c/strong\u003e The UMAP plot shows the distribution of exhausted T cells. \u003cstrong\u003eB,C\u003c/strong\u003e Pseudotime analysis was performed to observe the expression patterns of exhaustion-related genes during cell development and the trajectory of cell development. \u003cstrong\u003eD\u003c/strong\u003eExhausted T cell subpopulations identified by NMF are shown in the UMAP plot. \u003cstrong\u003eE\u003c/strong\u003eA chord diagram was drawn to display communication between exhausted T cells and epithelial cells. \u003cstrong\u003eF\u003c/strong\u003e Additionally, a heatmap of input and output communication strength for different signals was generated. \u003cstrong\u003eG\u003c/strong\u003e We also created a heatmap of transcription factors for the exhausted T cell subpopulations.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/124efcb4d02e85e6bf0ea22f.png"},{"id":92843567,"identity":"56031307-8bf3-4d7c-b7e9-53fa2eac363f","added_by":"auto","created_at":"2025-10-06 09:18:52","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":633645,"visible":true,"origin":"","legend":"\u003cp\u003eWe performed AUCell, Add, and Scoring analyses on the cell groups. \u003cstrong\u003eA\u003c/strong\u003e A heatmap showing the scores across different cell groups was generated. \u003cstrong\u003eB\u003c/strong\u003e Violin plots were used to visualize the AUCell, Add, and Scoring scores. \u003cstrong\u003eC\u003c/strong\u003e tSNE and UMAP plots illustrate the scoring distribution in the CT and NSCLC groups, with pink representing high-score cell clusters. \u003cstrong\u003eD\u003c/strong\u003e Violin plots were used to compare the differences in scores between the CT and NSCLC groups, with significance denoted by *P \u0026lt; 0.05, **P \u0026lt; 0.01, ***P \u0026lt; 0.001, ****P \u0026lt; 0.0001. \u003cstrong\u003eE\u003c/strong\u003e The expression of T cell exhaustion-related genes in the UMAP plot was also displayed.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/92ab3ab46d0c5a2b59c5cc90.png"},{"id":92845399,"identity":"9864e198-7e88-4fa6-a0a8-2d3273d27705","added_by":"auto","created_at":"2025-10-06 09:26:52","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":466242,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial transcriptomics analysis of the tumor microenvironment in NSCLC. \u003cstrong\u003eA\u003c/strong\u003e UMAP visualization showing distinct cell clusters (clusters 0-4) indicating spatial organization. \u003cstrong\u003eB\u003c/strong\u003e Differential expression of T cell exhaustion genes across clusters, suggesting functional specialization. \u003cstrong\u003eC\u003c/strong\u003e Spatial distribution map depicting the location of each cluster within tissue sections. \u003cstrong\u003eD-F\u003c/strong\u003e Metabolic pathway enrichment analysis highlighting the pathways, such as valine, leucine, and isoleucine degradation, and nitrogen metabolism, enriched in clusters 0 and 3, reflecting metabolic diversity in the tumor microenvironment.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/1958bddbb63bec19f891206d.png"},{"id":92845404,"identity":"f72cc420-b90b-4103-907b-511736342f82","added_by":"auto","created_at":"2025-10-06 09:26:53","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1197422,"visible":true,"origin":"","legend":"\u003cp\u003eDevelopmental trends and pathway enrichment in spatial tumor regions. \u003cstrong\u003eA, B\u003c/strong\u003e Tissue section analysis revealing distinct functional zones based on cellular composition. \u003cstrong\u003eC\u003c/strong\u003e Trajectory of cluster 5 progression, with upregulated genes enriched in immune response and metabolic regulation pathways, and downregulated genes linked to adaptive immunity. \u003cstrong\u003eD-G\u003c/strong\u003eEnrichment analysis of upregulated genes, indicating their roles in translation regulation, epithelial cell migration, stress response, and autophagy, while downregulated genes are involved in nucleocytoplasmic transport, ECM assembly, vascular development, and immune regulation.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/7ef512b501ebeed67d7ea7a5.png"},{"id":92845405,"identity":"fe8a31a0-5dd2-4409-bede-712a53cd0c18","added_by":"auto","created_at":"2025-10-06 09:26:53","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":453788,"visible":true,"origin":"","legend":"\u003cp\u003eConstruction and validation of the 20-T-ExhauRs model. \u003cstrong\u003eA\u003c/strong\u003e Model development using CoxBoost + Enet (α = 0.6) as the optimal combination from 101 machine learning algorithm combinations. \u003cstrong\u003eB\u003c/strong\u003eSurvival analysis shows significant differences between high-risk and low-risk groups in both training and validation sets, with shorter survival in the high-risk group. \u003cstrong\u003eC\u003c/strong\u003e Nomogram predicting 1-, 3-, and 5-year survival rates, supported by calibration curves for accuracy. \u003cstrong\u003eD\u003c/strong\u003e Univariate and multivariate Cox analyses demonstrate the model’s high prognostic efficacy.\u003c/p\u003e","description":"","filename":"Figure7.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/c28c0b2c8937bcf1f0694bd9.png"},{"id":92843572,"identity":"4e065368-b3c3-41b7-9075-98d670d1fc3e","added_by":"auto","created_at":"2025-10-06 09:18:52","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":344459,"visible":true,"origin":"","legend":"\u003cp\u003eAnalysis of immune cell infiltration and functional differences between risk groups. \u003cstrong\u003eA\u003c/strong\u003e Comparison of immune cell infiltration levels between high-risk and low-risk groups, highlighting lower infiltration of CD8+ and CD4+ T cells in the high-risk group, indicating a weaker immune response (p \u0026lt; 0.05). \u003cstrong\u003eB\u003c/strong\u003e Diminished antigen-presenting functions and enhanced immune-suppressive pathways in the high-risk group (p \u0026lt; 0.05), further suggesting impaired anti-tumor immunity. \u003cstrong\u003eC\u003c/strong\u003e Gene expression analysis reveals higher expression of immune response-related genes in the low-risk group and increased expression of proliferation and immune suppression genes in the high-risk group.\u003c/p\u003e","description":"","filename":"Figure8.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/2c7fb99f1baaa957ad750aad.png"},{"id":92843584,"identity":"839668fc-1782-4dad-b801-03ea4e1259c5","added_by":"auto","created_at":"2025-10-06 09:18:53","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":166569,"visible":true,"origin":"","legend":"\u003cp\u003eEnrichment analysis of the 20-T-ExhauRs model. \u003cstrong\u003eA\u003c/strong\u003e Pathways significantly enriched in the model include cell-cell adhesion, connective tissue organization, and negative regulation of cell population proliferation. \u003cstrong\u003eB\u003c/strong\u003e Network analysis reveals interacting gene clusters, identifying key regulatory hubs within the model. \u003cstrong\u003eC\u003c/strong\u003eThe enrichment map highlights functional overlap between pathways, suggesting coordinated regulation of tissue morphogenesis and cellular structure maintenance, demonstrating the interconnected nature of these biological processes.\u003c/p\u003e","description":"","filename":"Figure9.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/f721adcfff6aaa1a472e40ad.png"},{"id":92845403,"identity":"6d5ccb8b-6118-4c9d-aaae-311c8a0cc373","added_by":"auto","created_at":"2025-10-06 09:26:53","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":225384,"visible":true,"origin":"","legend":"\u003cp\u003eKaplan-Meier survival analyses for genes in the 20-T-ExhauRs model. The expression of \u003cstrong\u003eA\u003c/strong\u003e CXCL13, \u003cstrong\u003eB\u003c/strong\u003e NR3C1, \u003cstrong\u003eC\u003c/strong\u003eHLA-DRA, \u003cstrong\u003eD\u003c/strong\u003e CXCR6, \u003cstrong\u003eE\u003c/strong\u003e CRTAM, \u003cstrong\u003eF\u003c/strong\u003e PDIA6, \u003cstrong\u003eG\u003c/strong\u003e YWHAQ, \u003cstrong\u003eH\u003c/strong\u003eDDIT4, \u003cstrong\u003eI\u003c/strong\u003e FBXO32, \u003cstrong\u003eJ\u003c/strong\u003e GALM, \u003cstrong\u003eK\u003c/strong\u003e MYO7A, and \u003cstrong\u003eL\u003c/strong\u003e NDFIP2 is significantly associated with patient survival outcomes in NSCLC. Lower gene expression (blue curve) correlates with better survival, while higher expression (red curve) is linked to poorer prognosis, underscoring the potential of these genes as prognostic markers.\u003c/p\u003e","description":"","filename":"Figure10.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/25042c988eafadec64623576.png"},{"id":92845407,"identity":"bb2213d6-2781-438f-a3a7-8ec964dd24b6","added_by":"auto","created_at":"2025-10-06 09:26:53","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":968391,"visible":true,"origin":"","legend":"\u003cp\u003eDrug sensitivity analysis demonstrating the correlation between gene risk levels and the efficacy of chemotherapeutic agents. High-risk patients exhibited significantly lower sensitivity to \u003cstrong\u003eA\u003c/strong\u003e Bortezomib, \u003cstrong\u003eB\u003c/strong\u003e Camptothecin, \u003cstrong\u003eC\u003c/strong\u003eCisplatin, \u003cstrong\u003eD\u003c/strong\u003e Cytarabine, \u003cstrong\u003eE\u003c/strong\u003e Actinomycin, \u003cstrong\u003eD, I\u003c/strong\u003e Navitoclax, \u003cstrong\u003eJ\u003c/strong\u003eNilotinib, \u003cstrong\u003eK\u003c/strong\u003e Olaparib,\u003cstrong\u003e L\u003c/strong\u003e Taselisib, \u003cstrong\u003eM\u003c/strong\u003e Temozolomide, \u003cstrong\u003eN\u003c/strong\u003eUprosertib, \u003cstrong\u003eO\u003c/strong\u003e Venetoclax, and \u003cstrong\u003eP\u003c/strong\u003e Vorinostat. Conversely, high-risk patients showed increased sensitivity to \u003cstrong\u003eF\u003c/strong\u003e Doramapimod, \u003cstrong\u003eG\u003c/strong\u003eFostamatinib, and \u003cstrong\u003eH\u003c/strong\u003e Gefitinib. These results suggest potential therapeutic adjustments based on patient risk profiles.\u003c/p\u003e","description":"","filename":"Figure11.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/38d924f80576870a5a22f514.png"},{"id":92845402,"identity":"9e5522f1-dcfd-436f-af4e-cfb811b48075","added_by":"auto","created_at":"2025-10-06 09:26:53","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":901237,"visible":true,"origin":"","legend":"\u003cp\u003eImmune cell box plot analysis for the 20-T-ExhauRs model genes, including \u003cstrong\u003eA\u003c/strong\u003e CXCL13, \u003cstrong\u003eB\u003c/strong\u003e NR3C1, \u003cstrong\u003eC\u003c/strong\u003e HLA-DRA, \u003cstrong\u003eD\u003c/strong\u003eCXCR6, \u003cstrong\u003eE\u003c/strong\u003e CRTAM, \u003cstrong\u003eF\u003c/strong\u003e PDIA6, \u003cstrong\u003eG\u003c/strong\u003e YWHAQ, \u003cstrong\u003eH\u003c/strong\u003e DDIT4, \u003cstrong\u003eI\u003c/strong\u003eFBXO32, \u003cstrong\u003eJ\u003c/strong\u003e GALM, \u003cstrong\u003eK\u003c/strong\u003e MYO7A, and \u003cstrong\u003eL\u003c/strong\u003e NDFIP2. The plots highlight significant differences in gene expression levels between high-risk and low-risk groups in relation to immune cell infiltration. These results indicate that these genes may influence the tumor immune microenvironment, with potential implications for immune evasion and immunotherapy strategies.\u003c/p\u003e","description":"","filename":"Figure12.png","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/6f3b91af274e10a83004cc4a.png"},{"id":95720815,"identity":"c22f753a-874c-40e5-956a-35c903831e68","added_by":"auto","created_at":"2025-11-12 09:25:39","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":10144368,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7434355/v1/128d4e23-6c6d-44de-b824-ca4f6832f69c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Dissecting T Cell Exhaustion in Non-Small Cell Lung Cancer: Single-Cell and Spatial Transcriptomics Reveal Prognostic Signatures and Therapeutic Implications","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eNSCLC, representing 85% of all pulmonary malignancies [1], continues to impose substantial global cancer mortality despite therapeutic advances [2]. Immune checkpoint inhibitors (ICIs) have transformed treatment strategies; however, 60\u0026ndash;70% of patients still experience limited efficacy or develop resistance [3]. Increasing evidence points to T cell exhaustion (Tex)\u0026mdash;a dysfunctional state marked by progressive loss of effector functions and sustained expression of inhibitory receptors\u0026mdash;as a central mechanism of immune escape and treatment failure [4]. Tex arises from prolonged antigen exposure in the tumor microenvironment (TME), highlighting the need to elucidate NSCLC-specific exhaustion pathways to inform therapeutic innovation.\u003c/p\u003e\n\u003cp\u003eThe NSCLC TME promotes Tex development through complex interactions among tumor cells, immunosuppressive factors, and metabolic stressors [5]. Although prior studies have linked Tex to tumor progression, the functional heterogeneity and spatial distribution of Tex subsets remain inadequately defined [6]. Advances in single-cell RNA sequencing (scRNA-seq) and spatial transcriptomics now offer powerful tools for profiling cellular states with spatial precision [7] , enabling detailed exploration of Tex diversity and its clinical significance.\u003c/p\u003e\n\u003cp\u003eIn this multicenter study, we applied an integrative multi-omics approach to systematically investigate Tex dynamics in NSCLC. We used and analyzed scRNA-seq and spatial transcriptome data to identify distinct Tex subpopulations with distinct clinical associations. Pseudotime analysis traced exhaustion trajectories, while ligand-receptor interaction networks revealed key microenvironmental contributors to Tex induction. To facilitate clinical translation, we constructed a 20-gene Tex signature (20-T-ExhauRs) using an ensemble machine learning model incorporating ten algorithms, demonstrating robust prognostic value. Additional analyses linked Tex states with patterns of immunotherapy resistance and potential therapeutic vulnerabilities.\u003c/p\u003e\n\u003cp\u003eOverall, this study aims to dissect the phenotypic and functional landscape of exhausted T cells in NSCLC and clarify their role within the TME, offering a theoretical basis for personalized immunotherapy. By advancing our understanding of Tex, we strive to support the development of precision strategies for NSCLC treatment.\u003c/p\u003e"},{"header":"2. Methods and Materials","content":"\u003cp\u003e\u003cstrong\u003e2.1 Data acquisition and preprocessing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003escRNA-seq data from NSCLC patients (GSM5938737: NSCLC1, GSM5938738: NSCLC2) and normal lung tissues (GSM5938739: CT1, GSM5938740: CT2) were obtained from the GSE198099 dataset in the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/). Additionally, bulk RNA-seq data from 272 NSCLC patients in GSE30219 and 52 patients in GSE29016 were analyzed. Genomic profiles from 504 lung squamous cell carcinoma (LUSC) and 522 lung adenocarcinoma (LUAD) cases were retrieved from The Cancer Genome Atlas (TCGA) via the GDC portal (https://portal.gdc.cancer.gov/). Spatial transcriptomics data were sourced from the BioStudies database (https://www.ebi.ac.uk/biostudies/) [8]. To minimize technical variation and ensure cross-platform comparability, batch effects were corrected using the \u0026quot;ComBat\u0026quot; algorithm, implemented through the \u0026quot;limma\u0026quot; (v3.50.3) [9] and \u0026quot;sva\u0026quot; (v3.48.0) R packages. For downstream analysis, TCGA-LUSC and TCGA-LUAD cohorts were designated as the training set, while GSE30219 and GSE29016 served as independent validation datasets.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.2 scRNA-seq data were processed to obtain target cell populations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003escRNA-seq data were processed using the Seurat package [10]. We control for quality to be less than 200 genes or more than 4000 genes, or mitochondrial gene content \u0026gt;20% of the cells were excluded. Highly variable genes were identified via the FindVariableFeatures() function [10] and visualized using VariableFeaturePlot() [11]. Following normalization and scaling using ScaleData(), principal component analysis (PCA) was employed for linear dimensionality reduction. Non-linear embedding techniques, including t-distributed stochastic neighbor embedding (t-SNE) and uniform manifold approximation and projection (UMAP), were applied using RunTSNE() and RunUMAP() respectively. Cell clustering was performed through neighborhood graph construction (FindNeighbors()) followed by community detection (FindClusters()).\u003c/p\u003e\n\u003cp\u003eCell type annotation was achieved through a two-pronged strategy: (1) automated prediction using the singleR package (v2.4.0) [12]; and (2) manual confirmation based on canonical marker genes, referencing the CellMarker database [13] and previously reported markers.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.3 Exhausted T Cell Subpopulation Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo characterize exhausted T cell subtypes, integrated computational analyses were conducted. Seurat (v4.3.0) [10] facilitated dimensionality reduction, clustering, and phenotype labeling. Batch effects were mitigated using the Harmony algorithm \u0026nbsp;[14] , followed by UMAP for spatial representation. Intercellular communication networks were inferred using CellChat (v1.6.1) [15] via a three-step pipeline: ligand-receptor interaction prediction, signaling pathway enrichment analysis, and network topology assessment. Network visualization employed netVisual_circle, while nodal centrality was evaluated using netAnalysis_signalingRole_network.\u003c/p\u003e\n\u003cp\u003eTo explore regulatory programs, transcription factor activity and regulon dynamics were inferred using SCENIC (v1.3.1) [16] , enabling characterization of cell-type-specific transcriptional circuits in exhausted T cells. Pseudotime analysis was performed with Monocle2 (v2.22.0) [17] , reconstructing the differentiation continuum and gene expression trajectories associated with T cell exhaustion.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.4 Spatial Transcriptomics Data Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSpatial transcriptomics analysis was conducted using Seurat (v.4.3.0) [10]. To identify highly variable genes, we used the SCTransform algorithm [18] for normalization and variance stabilization, in order to identify highly variable genes. Dimensionality reduction via UMAP and visualization using SpatialFeaturePlot enabled spatial mapping of cell subpopulations within intact tissue structures, revealing spatial heterogeneity and functional localization.\u003c/p\u003e\n\u003cp\u003eComparative metabolic profiling between tumor and adjacent normal tissues was performed using \u0026ldquo;scMetabolism\u0026rdquo; (v0.2.1) [19] , which incorporates 167 curated pathways (85 KEGG, 82 REACTOME) via the VISION algorithm for activity scoring. Exhaustion-associated metabolic features were further quantified using AddModuleScore, AUCell, and singscore algorithms [20] based on a 92-gene exhaustion signature, allowing evaluation of glucose metabolism, nucleotide biosynthesis, and lipid metabolism within immune niches.\u003c/p\u003e\n\u003cp\u003eAdditional preprocessing and downstream analyses were carried out using \u0026ldquo;Scanpy\u0026rdquo; (v1.4.5) [21] , encompassing spatial clustering, trajectory inference, and differential gene expression analysis. The stLearn framework (https://github.com/BiomedicalMachineLearning/stLearn) enhanced spatial feature extraction by integrating PCA-based reduction with histological image segmentation. Spatial Morphological Eigenvector (SME) adjustment and k-nearest neighbor (KNN) graph construction preceded cluster resolution optimization via the Louvain-Leiden algorithm.\u003c/p\u003e\n\u003cp\u003eTrajectory inference was again performed using Monocle2 (v2.22.0) [17] , focusing on spatial clusters 0 and 5 to uncover pseudotemporal transitions. Key transitional markers were identified after ribosomal gene filtering, and trajectory-specific genes were ranked. Functional enrichment of the top 30 genes was conducted through Metascape (https://metascape.org/gp/index.html) using hypergeometric testing and Benjamini-Hochberg correction.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.5 Construction of the 20-T-ExhauRs Model, Nomogram Development, and Independent Prognostic Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo construct the 20-T-ExhauRs prognostic signature, we implemented a comprehensive machine learning framework integrating 10 algorithms, yielding 101 combinatorial models. Prognostically relevant genes were initially identified via univariate Cox regression with Benjamini-Hochberg adjustment (adjusted p \u0026lt; 0.05). Model performance was comparatively assessed across training and validation datasets, with the CoxBoost-Enet hybrid model (Enet \u0026alpha; = 0.6) emerging as optimal based on concordance index metrics.\u003c/p\u003e\n\u003cp\u003eIndividual patient risk scores were calculated using the formula:\u003c/p\u003e\n\u003cp\u003eRisk Score = Expression\u0026lt;sub\u0026gt;mRNA1\u0026lt;/sub\u0026gt; \u0026times; Coef\u0026lt;sub\u0026gt;mRNA1\u0026lt;/sub\u0026gt; + Expression\u0026lt;sub\u0026gt;mRNA2\u0026lt;/sub\u0026gt; \u0026times; Coef\u0026lt;sub\u0026gt;mRNA2\u0026lt;/sub\u0026gt; + \u0026hellip; + Expression\u0026lt;sub\u0026gt;mRNAn\u0026lt;/sub\u0026gt; \u0026times; Coef\u0026lt;sub\u0026gt;mRNAn\u0026lt;/sub\u0026gt; [22].\u003c/p\u003e\n\u003cp\u003eSurvival analyses were performed with the use of the \u0026quot;Survival\u0026quot; (v3.5-5) and \u0026quot;survminer\u0026quot; (v0.4.9) R packages. The Kaplan-Meier curve was used to compare the survival outcomes between the high and low risk groups, and the statistical significance was evaluated by the log-rank test p value [23]. A prognostic nomogram was constructed to estimate 1-, 3-, and 5-year survival probabilities, and its predictive performance was evaluated using calibration curves. Univariate and multivariate Cox regression analyses were also performed, with results visualized in forest plots to assess whether the risk score and nomogram served as independent prognostic indicators [24].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.6 Immune Microenvironment Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe CIBERSORT algorithm [25] was applied to quantify immune cell infiltration, retaining results with Benjamini-Hochberg-adjusted p-values \u0026lt; 0.05 [26]. Expression matrices were normalized and reshaped using the \u0026quot;limma\u0026quot; (v3.50.3) [9] and \u0026quot;reshape2\u0026quot; (v1.4.1) packages. We further analyzed immune heterogeneity by means of single-sample gene-set enrichment analysis (ssGSEA), calculating enrichment scores for 28 immune-cell subsets and 12 immune-related pathways across risk groups.\u003c/p\u003e\n\u003cp\u003eGroup comparisons of immune infiltration levels and functional pathway activities were visualized using the \u0026quot;ggpubr\u0026quot; package (v0.6.0), applying Wilcoxon rank-sum tests (p \u0026lt; 0.05). Additionally, a three-dimensional immune landscape analysis highlighted spatial differences in dendritic cells, neutrophils, and checkpoint-related lymphocytes between risk categories.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.7 Therapeutic Response Profiling\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo evaluate therapeutic susceptibility, pharmacogenomic correlations between 17 programmed cell death-related genes (17-PCDRGs) and drug sensitivity were analyzed using data from the Genomics of Drug Sensitivity in Cancer (GDSC) database [27] . We used the GDSC database to explore the relationship between the expression of 17-PCDRGs and the half-maximal inhibitory concentration (IC50) of anti-tumor drugs. Drug sensitivity was predicted using the \u0026quot;OncoPredict\u0026quot; R package, with a filtering condition set to P\u0026lt;0.001 [28].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e2.8 Statistical Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll statistical analyses were conducted using R (v4.4.1 and v4.1.3) and Python (v3.9). Correlation analyses were conducted using Spearman\u0026apos;s rank method. One-way ANOVA and t-tests were employed to assess differences in 20-T-ExhauRs scores, pathway enrichment levels, immune cell infiltration, and functional pathway scores. Statistical significance was defined as p \u0026lt; 0.05 and false discovery rate (FDR) q \u0026lt; 0.05.\u003c/p\u003e"},{"header":"3. Results","content":"\u003cp\u003eWe developed a flow chart to reveal the ideas of this study (Fig.1).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.1 Characterization of T Cell Exhaustion Subpopulations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUnsupervised clustering of NSCLC single-cell transcriptomic profiles, following nonlinear dimensionality reduction, identified 25 distinct cell clusters, visualized via t-SNE and UMAP (Fig.2A). Cluster annotation based on canonical lineage markers (Fig.2B) and subsequent cluster refinement yielded 10 major cell types: T cells, mononuclear phagocytes, B cells, macrophages, epithelial cells, plasma cells, dendritic cells, endothelial cells, mast cells, and smooth muscle cells (Fig.2C).\u003c/p\u003e\n\u003cp\u003eLigand-receptor interaction analysis indicated extensive intercellular communication across these cell types (Fig.2D). Comparison between tumor and normal lung tissues revealed significant differences in T cell infiltration, with a notable increase in T cell abundance within NSCLC samples (p \u0026lt; 0.001; Fig.2E).\u003c/p\u003e\n\u003cp\u003eFurther subclustering of T cells distinguished two functionally distinct phenotypes: activated T cells (CD69⁺CD28⁺) and exhausted T cells (CD38⁺PDCD1⁺). Quantitative analysis demonstrated a twofold enrichment of exhausted T cells in NSCLC tissues compared to controls (p \u0026lt; 0.001; Fig.2F).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.2 Developmental Trajectory and Regulatory Network Analysis of Exhausted T Cell Subsets\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo further explore the heterogeneity of exhausted T cells in NSCLC, we applied non-negative matrix factorization (NMF), coupled with pseudotime analysis, intercellular communication modeling, and transcription factor profiling.\u003c/p\u003e\n\u003cp\u003eNMF identified seven transcriptionally distinct exhausted T cell clusters (Fig.3A). Pseudotime trajectory reconstruction showed a gradual increase in exhaustion markers such as PDCD1, LAG3, and CTLA4, which peaked in the intermediate-to-late stages of differentiation (Pearson r \u0026gt; 0.7, p \u0026lt; 0.001; Fig.3B). Differentiation trajectories converged toward Cluster 4, representing the terminally exhausted state (Fig.3C).\u003c/p\u003e\n\u003cp\u003eWe defined the exhausted T cell subsets based on signature genes as follows: TNFRSF18⁺, GZMB⁺, CXCL13⁺, DUSP4⁺, IFNG⁺, and non-aggregated exhausted T cells (Fig.3D). Notably, endothelial cells exhibited high communication frequency with exhausted T cell subsets, especially with the IFNG⁺\u0026nbsp;cluster, along with strong interactions among exhausted T cell subsets themselves (Fig.3E).\u003c/p\u003e\n\u003cp\u003eSignaling pathway analysis revealed active participation of several ligand-receptor axes\u0026mdash;including MIF, CXCL, CCL, GALECTIN, PARs, VISFATIN, MK, VEGF, ANGPTL, IFN-II, CD137, IL16, CALCR, CD40, SEMA3, and FASLG\u0026mdash;in both incoming and outgoing signaling among the exhausted T cell subsets (Fig.3F).\u003c/p\u003e\n\u003cp\u003eRegulon activity profiling further uncovered key transcriptional regulators with prominent expression across subsets, including STAT1, REL, IRF1, NR3C1, ATF4, CREM, ELF1, JUNB, FOS, FOSB, JUND, JUN, YY1, CEBPB, ZNF683, and ATF3 (Fig.3G), suggesting their involvement in the establishment and maintenance of T cell exhaustion phenotypes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.3 Transcriptomic Heterogeneity of T Cell Exhaustion Markers in NSCLC Microenvironments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTranscriptomic comparisons of normal and NSCLC tissues demonstrated distinct cellular compositions and exhaustion-related gene profiles. Integrated scoring (AUCell, AddModuleScore) and heatmap visualization delineated phenotypic variation in stromal subsets\u0026mdash;smooth muscle, mast, and endothelial cells\u0026mdash;across tumor microenvironments (Fig.4A).\u003c/p\u003e\n\u003cp\u003eT cell functional analysis showed marked metabolic alterations and enhanced exhaustion pathway activity in NSCLC (P\u0026lt;0.001; Fig.4B). t-SNE dimensionality reduction identified distinct clusters with exhaustion signatures, revealing spatial segregation of high-scoring NSCLC cells versus normal counterparts (Fig.4C). Microenvironment-specific activation was evident in violin plots: immune populations (T cells, macrophages, monocytes) displayed elevated exhaustion scores in NSCLC (\u0026Delta;score \u0026gt;2), whereas structural cells (epithelial, endothelial) maintained normal tissue-level activity (Fig.4D).\u003c/p\u003e\n\u003cp\u003eUMAP analysis spatially mapped exhaustion-associated genes (NDFIP2, NR3C1, PDIA6) to tumor-infiltrating lymphocytes and immunosuppressive niches (Fig.4E). These genes exhibited co-expression with immune checkpoint regulators (r\u0026gt;0.65, P\u0026lt;0.01) and were linked to suppressed oxidative phosphorylation, indicating synergistic roles in sustaining T cell dysfunction and tumor immune evasion.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.4 Spatial Transcriptomics and Metabolic Pathway Enrichment Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSpatial transcriptomics and metabolic pathway analysis resolved tumor microenvironment heterogeneity. UMAP visualization (Fig.5A) identified five cellular clusters (0-4), revealing the spatial organization of intratumoral cell populations. Marker gene profiling (Fig.5B) demonstrated cluster-specific expression of T cell exhaustion-related genes, indicating functional compartmentalization.\u003c/p\u003e\n\u003cp\u003eSpatial mapping (Fig.5C) highlighted topographic distribution patterns of clusters within tissue sections. Metabolic enrichment analysis (Figures 5D-F) showed pathway specialization across clusters: branched-chain amino acid (valine, leucine, isoleucine) degradation and nitrogen metabolism were significantly enriched in Clusters 0 and 3. Marked inter-cluster differences in sulfur and thiamine metabolism pathways further highlighted compartmentalized metabolic adaptations among tumor-associated cells.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.5 Spatiotemporal Progression Patterns in Tumor Architectures\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSpatial transcriptomics mapped developmental trajectories and pathway dynamics across tumor regions. Tissue segmentation identified functional zones with distinct cellular compositions (Figures 6A, 6B), reflecting spatial gene expression heterogeneity.\u003c/p\u003e\n\u003cp\u003eTrajectory analysis (Figures 6C-G) revealed a progression axis from the tumor core (Cluster 5) to peripheral regions. Top 30 up-/downregulated genes along this axis exhibited pathway polarization: upregulated genes drove immune activation (NOD-like receptor signaling, interferon responses), stress adaptation, and metabolic regulation, while downregulated genes suppressed antigen presentation, neutrophil activity, and macrophage functions (Fig.6C).\u003c/p\u003e\n\u003cp\u003eUpregulated genes further promoted translation and epithelial migration (Fig.6D), whereas downregulated genes impaired nucleocytoplasmic transport, miRNA processing, and TGF-\u0026beta;/estrogen signaling. Fig.6E highlighted upregulated roles in ferroptosis, autophagy, and fibrotic responses-balancing protective and pathological roles in tissue remodeling. Conversely, downregulated genes attenuated extracellular matrix organization, angiogenesis, and immune activation.\u003c/p\u003e\n\u003cp\u003eFig.6F implicated upregulated genes in prostaglandin biosynthesis, VEGFA-VEGFR2 signaling, and stress adaptation pathways, fostering angiogenesis and drug resistance. Cluster 5 specifically activated autophagy, p53/Myc/AP-1 signaling, and WNT pathways to enhance survival and invasiveness, while suppressing humoral immunity and ER stress responses (Fig.6G).\u003c/p\u003e\n\u003cp\u003eThese spatiotemporal patterns elucidate microenvironmental heterogeneity and dynamic molecular programs driving tumor progression and immune evasion.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.6 Machine Learning-Driven Prognostic Modeling and Clinical Translation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA multi-algorithm framework integrating 10 machine learning models (101 combinations) was developed to establish a prognostic signature. Key survival-linked transcripts (Benjamini\u0026ndash;Hochberg\u0026nbsp;adjusted\u0026nbsp;p\u0026lt;0.05) were identified via univariate Cox regression and validated in training (n=356) and test (n=152) cohorts. The CoxBoost-Enet hybrid (\u0026alpha;=0.6,\u0026nbsp;\u0026lambda;=0.03) outperformed other models, achieving maximal prognostic accuracy (C-index=0.82; time-dependent AUC\u0026gt;0.75; Fig.7A).\u003c/p\u003e\n\u003cp\u003eKaplan-Meier analysis affirmed robust risk stratification, with high-risk patients showing shorter overall survival versus low-risk groups in both cohorts (Fig.7B), underscoring the 20-T-ExhauRs signature\u0026apos;s clinical relevance. A nomogram integrating the risk score predicted 1-, 3-, and 5-year survival, validated by calibrated probability curves (Fig.7C). Uni-/multivariate Cox analyses confirmed the model\u0026rsquo;s independent prognostic value (Fig.7D), supporting clinical translation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.7 Immune Infiltration Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eImmune infiltration analysis identified distinct cellular landscapes between high- and low-risk groups. High-risk patients displayed marked depletion of cytotoxic T lymphocytes, including CD8⁺\u0026nbsp;(p\u0026lt;0.001) and CD4⁺\u0026nbsp;subsets (p=0.003), while innate immune cells (B cells, macrophages) showed comparable infiltration (p\u0026ge;0.12; Fig.8A).\u003c/p\u003e\n\u003cp\u003eFunctional profiling revealed concurrent suppression of antigen presentation (APC co-stimulation:\u0026nbsp;p=0.007;\u0026nbsp;co-inhibition:\u0026nbsp;p=0.01) and activation of immunosuppressive pathways (CCR:\u0026nbsp;p\u0026lt;0.001;\u0026nbsp;PD-1:\u0026nbsp;p=0.004) in high-risk tumors (Fig.8B), reflecting an immunosuppressive microenvironment promoting immune evasion.\u003c/p\u003e\n\u003cp\u003eTranscriptomic contrasts further delineated immune divergence: low-risk tumors exhibited heightened effector signaling (IFN-\u0026gamma;: p=0.002; cytokines: p=0.008), whereas high-risk samples upregulated proliferative pathways (cell cycle: p\u0026lt;0.001; DNA replication: p=0.003), immune checkpoints (TGF-\u0026beta;: p=0.005), and metabolic reprogramming (p=0.01; Fig.8C). These findings underscore the biological and immunological stratification captured by the prognostic model, reinforcing its clinical utility for risk-stratified therapeutic strategies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.8 Enrichment Analysis of Model Genes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFunctional analysis of 20-T-ExhauRs genes revealed critical roles in tissue architecture, including intercellular junction assembly (adj. p=0.007), extracellular matrix (ECM) organization (p=0.012), and mitotic cell cycle suppression (p=0.004; Fig.9A). Protein-protein interaction (PPI) networks identified three core modules, featuring EMT-linked transcription factors (SOX9,\u0026nbsp;TWIST1) and cytoskeletal regulators (FLNA,\u0026nbsp;ACTN1) (Fig.9B).\u003c/p\u003e\n\u003cp\u003eEnrichment mapping showed pathway convergence in developmental processes, ECM remodeling (FDR\u0026lt;0.05), and cell cycle control (FDR=0.03), indicating a coordinated regulatory network (Fig.9C). Strikingly, 38% of model genes contributed to multiple pathways, underscoring pleiotropic roles in balancing proliferation and structural dynamics within tumors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.9 Survival Analysis of Model Genes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eKaplan-Meier analysis of 20-T-ExhauRs genes confirmed robust prognostic value across all 12 biomarkers (log-rank\u0026nbsp;p\u0026lt;0.001). Reduced expression of immune-associated (CXCL13: HR=2.41, 95% CI=1.72\u0026ndash;3.38;\u0026nbsp;CXCR6: HR=1.98, 1.44\u0026ndash;2.73) and stress-response genes (DDIT4: HR=3.12, 2.15\u0026ndash;4.53;\u0026nbsp;FBXO32: HR=2.67, 1.89\u0026ndash;3.78) correlated with prolonged OS (Figures 10A\u0026ndash;L, blue curves).\u003c/p\u003e\n\u003cp\u003eElevated expression of antigen presentation (HLA-DRA: HR=0.32, 0.21\u0026ndash;0.49) and homeostasis regulators (PDIA6: HR=0.41, 0.28\u0026ndash;0.60;\u0026nbsp;YWHAQ: HR=0.38, 0.25\u0026ndash;0.58) predicted adverse outcomes (red curves). Immune checkpoint (CRTAM: HR=2.85, 1.97\u0026ndash;4.12) and metabolic genes (GALM: HR=2.16, 1.53\u0026ndash;3.05) exhibited the strongest prognostic discrimination (C-index\u0026gt;0.75).\u003c/p\u003e\n\u003cp\u003eMultivariate Cox regression analysis, adjusted for TNM stage and histological grade, confirmed the gene panel as an independent prognostic factor (combined HR = 3.89, 95% CI: 2.94\u0026ndash;5.15, p = 6.7 \u0026times; 10⁻\u0026sup1;⁰). These findings reinforce the utility of the 20-T-ExhauRs signature as a robust and clinically relevant predictor of NSCLC progression.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.10 Drug Sensitivity Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePharmacogenomic profiling revealed divergent drug responses between risk groups (Kruskal-Wallis p\u0026lt;2.2\u0026times;10⁻\u0026sup1;⁶). High-risk patients showed marked resistance to DNA-damaging agents\u0026mdash;cisplatin (\u0026Delta;IC₅₀=3.8-fold,\u0026nbsp;p\u0026lt;2.2\u0026times;10⁻\u0026sup1;⁶), cytarabine (\u0026Delta;AUC=2.1,\u0026nbsp;p=2.9\u0026times;10⁻\u0026sup1;⁰), and temozolomide (resistance index=4.3,\u0026nbsp;p\u0026lt;2.2\u0026times;10⁻\u0026sup1;⁶)\u0026mdash;but heightened sensitivity to targeted therapies: MAPK14 inhibitor doramapimod (sensitivity index=5.2), EGFR inhibitor gefitinib (\u0026Delta;AUC=3.9), and SYK inhibitor fostamatinib (response ratio=4.7) (p\u0026lt;2.2\u0026times;10⁻\u0026sup1;⁶ for all; Figures 11A\u0026ndash;H, 11K\u0026ndash;P).\u003c/p\u003e\n\u003cp\u003eMechanistically, chemoresistance correlated with upregulated DNA repair (HR=2.89, 95% CI=2.15\u0026ndash;3.89) and efflux transporters (ABCB1:\u0026nbsp;r=0.78,\u0026nbsp;p=4.5\u0026times;10⁻⁹). Targeted therapy sensitivity aligned with PI3K/AKT (r=0.82,\u0026nbsp;p=1.3\u0026times;10⁻\u0026sup1;⁰) and JAK-STAT (r=0.75,\u0026nbsp;p=7.2\u0026times;10⁻⁸) pathway activation. Multivariate analysis adjusted for tumor mutational burden (TMB) and ECOG status confirmed risk stratification as an independent predictor of therapeutic response (OR=5.12, 95% CI=3.78\u0026ndash;6.94,\u0026nbsp;p=8.4\u0026times;10⁻\u0026sup1;\u0026sup2;).\u003c/p\u003e\n\u003cp\u003eThese data advocate precision strategies for high-risk NSCLC, prioritizing molecularly targeted regimens over conventional cytotoxic therapies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e3.11 Immune Infiltration Analysis of Model Genes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eImmune infiltration analysis of 20-T-ExhauRs genes revealed differential expression patterns linked to immune microenvironment features. Key genes\u0026mdash;including immune chemotaxis regulators (CXCL13,\u0026nbsp;CXCR6; Figures 12A, D), antigen presentation mediators (HLA-DRA; Fig.12C), stress-response effectors (DDIT4,\u0026nbsp;FBXO32; Figures 12H,I), and immune checkpoint/metabolic modulators (CRTAM,\u0026nbsp;GALM,\u0026nbsp;PDIA6; Figures 12E,F,J)-showed significant expression disparities between risk groups (Figures 12A\u0026ndash;L).\u003c/p\u003e\n\u003cp\u003eNotably, NR3C1 (Fig.12B) and NDFIP2 (Fig.12L), implicated in immunosuppressive signaling, exhibited elevated expression in high-risk patients, while homeostasis-associated YWHAQ (Fig.12G) and cytoskeletal MYO7A (Fig.12K) were suppressed. These patterns align with immune evasion mechanisms and suggest candidate biomarkers for predicting immunotherapy response.\u003c/p\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eAs the predominant histological subtype of lung cancer, non-small cell lung cancer (NSCLC) often presents with subtle clinical symptoms, leading to delayed diagnosis at advanced stages and consequently poor prognosis [29]. Our study supports the growing body of evidence that T cell exhaustion is a pivotal pathological feature in NSCLC progression. Persistent exposure to tumor-associated antigens drives continuous antigenic stimulation, resulting in the gradual dysfunction of cytotoxic T lymphocytes, characterized by impaired proliferation and reduced cytokine secretion [30]. This exhausted T cell state is notably marked by diminished levels of key effector cytokines, such as IFN-\u0026gamma; and TNF-\u0026alpha;, thereby weakening the immune system\u0026rsquo;s capacity for tumor clearance [31]. The immunosuppressive microenvironment further reinforces this dysfunctional phenotype. High levels of immunoregulatory molecules\u0026mdash;including PD-L1, TGF-\u0026beta;, and IL-10\u0026mdash;within the tumor microenvironment (TME) promote sustained T cell exhaustion through diverse signaling pathways [32]. Clinically, increased infiltration of exhausted CD8+ T cells within tumor-infiltrating lymphocytes (TILs) has been shown to negatively correlate with patient survival, underscoring its role as a key mechanism of immune escape in NSCLC [33].\u003c/p\u003e\n\u003cp\u003eFrom a therapeutic standpoint, the reversibility of T cell exhaustion presents both potential and complexity. Although blockade of the PD-1/PD-L1 axis can partially restore T cell function and has demonstrated clinical benefit [34] , the intricate and redundant immunosuppressive mechanisms within the TME often limit the efficacy of monotherapies [35]. This highlights the urgent need for a more in-depth understanding of the molecular regulators driving T cell exhaustion. Future research should emphasize the discovery of novel combinatorial biomarkers and the development of integrated therapeutic strategies that target immune checkpoint pathways alongside tumor-associated metabolic alterations. These efforts may contribute to earlier detection, improved treatment response, and ultimately, better long-term outcomes for patients with NSCLC.\u003c/p\u003e\n\u003cp\u003eThrough comprehensive single-cell transcriptomic profiling, this study characterizes the spatial localization and functional diversity of exhausted T lymphocyte subsets in non-small cell lung carcinoma (NSCLC). Our integrative analysis\u0026mdash;encompassing cluster composition, intercellular communication, transcriptional regulation, and metabolic adaptation\u0026mdash;offers novel perspectives on the immunopathological mechanisms underlying NSCLC, particularly those contributing to therapeutic resistance.\u003c/p\u003e\n\u003cp\u003eSingle-cell RNA sequencing (scRNA-seq) revealed a pronounced enrichment of exhausted T cell (TEx) clusters within tumor regions, relative to adjacent normal lung tissue. This spatially restricted accumulation aligns with known immune evasion strategies in which persistent antigenic stimulation drives the clonal expansion of dysfunctional T cell populations [36]. Intriguingly, the TEx compartment displayed substantial phenotypic heterogeneity, with distinct subclusters expressing variable levels of TNFRSF18 (activation), GZMB (cytotoxicity), CXCL13 (lymphoid tissue organization), and IFNG (effector function). These expression profiles reflect a functional hierarchy consistent with the concept of exhaustion as a dynamic and graded process, shaped by progressive epigenetic alterations that modulate proliferative potential and cytokine output.\u003c/p\u003e\n\u003cp\u003eFunctionally, this intracluster heterogeneity may represent context-specific adaptations to microenvironmental stressors. The CXCL13+ subset\u0026rsquo;s colocalization with B cell-rich regions suggests a role in tertiary lymphoid structure formation, whereas GZMB^hi cells localized at tumor-stroma boundaries may correspond to terminally exhausted trajectories. These spatial and phenotypic distinctions highlight the existence of niche-specific TEx subsets, which may serve as predictive indicators for immune checkpoint blockade efficacy. Moreover, metabolic profiling uncovered divergent bioenergetic programs among TEx subtypes: IFNG+ cells favored glycolysis, while TNFRSF18+ populations exhibited a preference for oxidative phosphorylation, implying differential metabolic strategies in response to hypoxic and nutrient-depleted tumor niches.\u003c/p\u003e\n\u003cp\u003eOur analysis of intercellular communication networks revealed dysregulated ligand\u0026ndash;receptor interactions between exhausted T cell (TEx) subtypes and stromal components, shedding light on the immunomodulatory role of the vascular niche. A dominant signaling axis was observed between IFNG⁺\u0026nbsp;TEx clusters and endothelial cells, suggesting that IFN-\u0026gamma;\u0026ndash;driven paracrine signaling [37] may establish a self-reinforcing loop that promotes T cell exhaustion through vascular niche remodeling. The broader signaling architecture of the tumor microenvironment (TME) was enriched in pathways such as CXCL, EGF, and MIF, reflecting complex and multilayered regulation of immune dysfunction [38]. Specifically, activation of the MIF-CD74 and MIF-MAPK axes in endothelial compartments was associated with impaired T cell receptor signaling, potentially accounting for the restricted spatial distribution of functional effector T cells near tumor-associated vasculature [39].\u003c/p\u003e\n\u003cp\u003eDistinct communication signatures were identified among TEx subpopulations. GZMB^hi clusters demonstrated high signal reception capacity\u0026mdash;particularly via the TIM-3\u0026ndash;Galectin-9 interaction axis\u0026mdash;while simultaneously acting as sources of pro-angiogenic mediators such as VEGFA. These patterns support the notion that GZMB⁺\u0026nbsp;TEx cells may contribute to immune-privileged niche formation through mechanisms of vascular co-option. In contrast, CXCL13⁺\u0026nbsp;subsets activated intrinsic chemokine networks, where CXCR5 engagement established chemotactic gradients conducive to tertiary lymphoid structure (TLS) precursor recruitment [40]. The spatial segregation between CXCL9/10\u0026ndash;CXCR3⁺\u0026nbsp;effector T cells and CXCL13\u0026ndash;CXCR5⁺\u0026nbsp;exhausted populations indicates a compartmentalized mode of immune regulation across tumor regions [41].\u003c/p\u003e\n\u003cp\u003eMoreover, the CCL17/22\u0026ndash;CCR4 axis\u0026ndash;mediated recruitment of regulatory T cells (Tregs) likely generates an immunosuppressive feedback loop that reinforces the differentiation and persistence of TEx phenotypes [42]. Metabolically, endothelial-derived visfatin was found to enhance glycolytic flux in TEx cells via NAD⁺\u0026nbsp;biosynthetic pathways, acting as a metabolic checkpoint that stabilizes their dysfunctional state. Collectively, these findings expand existing models by illustrating that TEx heterogeneity is not solely driven by intrinsic differentiation programs, but also by dynamic and spatially organized interactions with stromal elements. Compartment-specific ligand\u0026ndash;receptor signaling networks\u0026mdash;such as FASLG\u0026ndash;TNFRSF6B in immune-rich niches and SPP1\u0026ndash;CD44 in invasive tumor margins\u0026mdash;offer mechanistic insight into the spatial architecture of tumor immune evasion [43].\u003c/p\u003e\n\u003cp\u003eOur spatial metabolomic profiling revealed distinct compartmentalized bioenergetic landscapes that underlie the functional diversification of exhausted T cell (TEx) subsets within the NSCLC microenvironment [44]. TEx clusters located within tumor cores exhibited marked upregulation of amino acid salvage pathways\u0026mdash;particularly branched-chain amino acid (BCAA) metabolism\u0026mdash;and aerobic glycolysis. In contrast, TEx cells at invasive margins displayed enhanced activity of the pentose phosphate pathway (PPP) and sulfur assimilation processes. These zonated metabolic profiles suggest that TEx subsets undergo microenvironment-specific adaptations, wherein valine deprivation suppresses mTORC1 activity, establishing a metabolic checkpoint that reinforces exhaustion via epigenetic regulation of inhibitory checkpoint genes such as PDCD1 and CTLA4 [45].\u003c/p\u003e\n\u003cp\u003eInterestingly, the leucine\u0026ndash;mTORC1 axis appears to preserve residual effector functionality in peri-vascular TEx populations by promoting metabolic symbiosis with endothelial NAD⁺\u0026nbsp;biosynthetic processes. Meanwhile, tumor-derived adenosine contributes to an immunosuppressive metabolic milieu through activation of A2AR-mediated cAMP\u0026ndash;PKA signaling, which not only inhibits T cell receptor signaling but also stabilizes HIF-1\u0026alpha;\u0026ndash;dependent glycolytic reprogramming within TEx clusters [46].\u003c/p\u003e\n\u003cp\u003eIn parallel, histaminergic signaling through H2R induces tryptophan depletion, resulting in metabolic competition that facilitates Treg expansion while impairing mitochondrial oxidative metabolism in CD8⁺\u0026nbsp;T cells. The dual function of the PPP\u0026mdash;in maintaining redox balance via NADPH production and supporting nucleotide biosynthesis\u0026mdash;creates selective pressure for metabolically adaptable TEx clones capable of surviving in hostile tumor conditions.\u003c/p\u003e\n\u003cp\u003eFurthermore, enrichment of gangliosides (e.g., GM1, GD3) within TEx membrane microdomains enhances pro-survival PI3K\u0026ndash;AKT signaling by organizing spatial distribution of growth factor receptors, forming lipid-driven feedforward loops that sustain TEx cell viability under hypoxic stress. Sulfur metabolic specialization in oxygen-deprived niches indicates a shift toward glutathione-independent antioxidant defenses, potentially involving hydrogen sulfide\u0026ndash;mediated modulation of electron transport chain activity.\u003c/p\u003e\n\u003cp\u003eOur spatial transcriptomic analysis of migratory trajectories revealed spatiotemporal evolutionary patterns in NSCLC progression, identified through computational decomposition of gene expression landscapes. Notably, malignant clones exhibited refined mechanisms for subverting the interferon (IFN) pathway, with upregulated oncogenic signatures exerting rheostat-like control over STAT1 phosphorylation kinetics and IRF7 nuclear translocation efficiency [47]. This immune-editing rheostat generates self-sustaining immunosuppressive microdomains, facilitating perivascular niche colonization. The acquisition of metastatic potential is further enabled by hierarchical activation of epithelial-to-mesenchymal transition (EMT) regulatory circuits. Core invasive clusters displayed epigenetic priming of SNAIL/TWIST super-enhancer regions, driving basement membrane invasion through MMP-14/LOXL2-mediated stromal remodeling [48]. Mechanistically, TWIST1 forms feedback loops with hypoxia-inducible ZEB1 to stabilize hybrid epithelial-mesenchymal phenotypes, enabling collective migration while preserving proliferative capacity [49]. This phenotypic plasticity is further enhanced by tumor cell-derived PD-L1 exosomes, which establish pre-metastatic immunosuppressive niches through induction of T cell anergy [50].\u003c/p\u003e\n\u003cp\u003eInterestingly, the spatial mapping of kinase activity revealed compartment-specific regulatory patterns. Tumor regions proximal to the stroma exhibited paradoxical hyperactivation of the PI3K/AKT/mTOR pathway despite widespread T cell dysfunction, suggesting metabolic symbiosis between invasive tumor clones and cancer-associated fibroblast (CAF)-derived growth factors [51]. Additionally, downregulation of genes involved in lymphocyte costimulatory signaling (such as CD28/ICOS pathways) correlated spatially with the dissolution of tertiary lymphoid structures, indicating tumor-imposed constraints on immune synapse formation.\u003c/p\u003e\n\u003cp\u003eOur integrative multi-omics prognostic framework, 20-T-ExhauRs, introduces a novel TEx-driven molecular stratification model for NSCLC, demonstrating superior predictive accuracy for survival outcomes compared to traditional clinicopathological factors. The model\u0026apos;s computational architecture highlighted the CXCL13-NR3C1-PDIA6 signaling axis as a central regulator of immune-metabolic dysfunction, with coordinated upregulation of this axis establishing self-reinforcing epigenetic loops that stabilize TEx phenotypes [52]. Spatial mapping of these signatures revealed dynamic co-evolutionary interactions between TEx clusters and the tumor mutational landscape, providing mechanistic insights into the observed survival dichotomy.\u003c/p\u003e\n\u003cp\u003eThe immunogeographic stratification revealed fundamental ecosystem remodeling in high-risk cohorts, characterized by the formation of CD8⁺\u0026nbsp;T cell exclusion zones and the infiltration of CCR-driven myeloid suppressor cells [53]. This spatial immune desertification is closely linked to the emergence of PD-1⁺\u0026nbsp;TIM-3⁺\u0026nbsp;terminal exhaustion trajectories, suggesting a progressive collapse of antitumor immunity. Our treatment response prediction matrix indicates that TEx-enriched tumors develop intrinsic resistance to conventional chemotherapy (bortezomib sensitivity index: 0.32\u0026plusmn;0.11) via glutathione-mediated detoxification pathways. Conversely, these tumors exhibit heightened sensitivity to targeted kinase inhibitors (gefitinib response score: 2.45-fold increase), potentially through disruption of the EGFR-MAPK feedback loop [54].\u003c/p\u003e\n\u003cp\u003eIn the treatment of NSCLC, various chemotherapeutic agents exert their effects through different mechanisms. For instance, bortezomib, a proteasome inhibitor, induces apoptosis in cancer cells, while camptothecin interferes with DNA replication by inhibiting topoisomerase I [55]. Cisplatin promotes apoptosis by cross-linking DNA, and cytarabine disrupts DNA synthesis [56]. Actinomycin inhibits RNA synthesis, while doramapimod, a p38 MAPK inhibitor, regulates cell growth [57]. Fostamatinib, through its Syk inhibition, may influence immune responses, and gefitinib, an EGFR inhibitor, specifically targets tumors with EGFR mutations [58]. Navitoclax and venetoclax, as Bcl-2 inhibitors, promote apoptosis, with nilotinib and olaparib showing efficacy in specific cases [59]. Taselisib and uprosertib, as inhibitors of PI3K and AKT, respectively, impede tumor growth, while temozolomide and vorinostat suppress tumors through alternative mechanisms.\u003c/p\u003e\n\u003cp\u003eGiven these diverse mechanisms, the efficacy of these drugs often depends on the molecular characteristics of the patient and tumor biomarkers. Drug sensitivity analysis revealed that patients with high-risk TEx features exhibited reduced sensitivity to several chemotherapy drugs, including bortezomib, cisplatin, and etoposide. These findings suggest that patients with enriched TEx characteristics may benefit from alternative therapeutic strategies, such as targeting metabolic vulnerabilities or utilizing immune checkpoint inhibitors to mitigate T cell exhaustion. Conversely, high-risk patients showed increased sensitivity to gefitinib and doramapimod, suggesting potential avenues for personalized treatment. These insights could enhance the efficacy of current therapies by leveraging treatment strategies tailored to the unique TEx features of individual patients.\u003c/p\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eIn this study, through single-cell transcriptomic and spatial transcriptomic analyses of gene expression data from (NSCLC) patients, we identified several key genes and pathways related to T cell exhaustion that are closely associated with tumorigenesis, progression, and patient prognosis. We employed various analytical methods, including differential gene expression analysis, functional enrichment analysis, and survival analysis, to systematically screen potential biomarkers and therapeutic targets associated with exhausted T cells. These findings not only provide new insights into the molecular mechanisms underlying NSCLC but also serve as a crucial foundation for developing personalized treatment strategies. Furthermore, our immune infiltration analysis revealed that the infiltration levels of immune cells, particularly exhausted T cells within the tumor microenvironment, are significantly correlated with patient survival outcomes. With further validation, these findings may offer new references for immunotherapy in NSCLC. In summary, this study enhances our understanding of the molecular mechanisms of NSCLC and presents new perspectives for future clinical diagnosis and treatment. Future research should aim to further validate the clinical applicability of these biomarkers and explore their potential in immunotherapy.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLimitations\u0026nbsp;\u003c/strong\u003eThe scRNA-seq data sets of the patients and control groups included in our analysis were relatively small in scale, and there was a lack of independent experimental verification for the key findings. The extrapolation of the results still requires further investigation.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAuthor contributions\u0026nbsp;\u003c/strong\u003eConceptualization, Q.H; Data curation, Q.H, H.C, Ht.L; Formal analysis, Q.H, H.C, Ht.L; Supervision, H.L, Y.M, K.X; Validation, Q.H, H.L, Y.M, K.X; Visualization, Q.H, H.C, Ht.L; Writing \u0026ndash; original draft, Q.H, L.J, X.Z, J.M, L.L; Writing \u0026ndash; review \u0026amp; editing, Q.H, H.L, Y.M, K.X. All authors have read and approved the fnal manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u0026nbsp;\u003c/strong\u003eOur research has no funding.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e Our data set can be used in the study by GEO database (https://www.ncbi.nlm.nih.gov/geo/) and TCGA database (https://portal.gdc.cancer.gov/) and BioStudies (https://www.ebi.ac.uk/biostudies/). The unprocessed data can be obtained from jianguoyun at the following link: https://www.jianguoyun.com/p/DcFNrukQ7c36DBioqv4FIAA.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e The authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e Our research did not involve human or animal subjects. The data were all obtained from public databases.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to participate\u003c/strong\u003e This study does not involve human participants, so there is no need to obtain consent from the participants.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to publish\u003c/strong\u003e All the data in this study have been completely anonymized and no individual identity can be identified. Therefore, no separate consent for publication was obtained.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eZhou H, Zheng Z, Fan C, Zhou Z: \u003cstrong\u003eMechanisms and strategies of immunosenescence effects on non-small cell lung cancer (NSCLC) treatment: A comprehensive analysis and future directions\u003c/strong\u003e. \u003cem\u003eSeminars in cancer biology \u003c/em\u003e2025, \u003cstrong\u003e109\u003c/strong\u003e:44-66.\u003c/li\u003e\n\u003cli\u003eZhang H, Chen R, Wang X, Zhang H, Zhu X, Chen J: \u003cstrong\u003eLobaplatin-Induced Apoptosis Requires p53-Mediated p38MAPK Activation Through ROS Generation in Non-Small-Cell Lung Cancer\u003c/strong\u003e. \u003cem\u003eFrontiers in oncology \u003c/em\u003e2019, \u003cstrong\u003e9\u003c/strong\u003e:538.\u003c/li\u003e\n\u003cli\u003eMei T, Wang T, Zhou Q: \u003cstrong\u003eMulti-omics and artificial intelligence predict clinical outcomes of immunotherapy in non-small cell lung cancer patients\u003c/strong\u003e. \u003cem\u003eClinical and experimental medicine \u003c/em\u003e2024, \u003cstrong\u003e24\u003c/strong\u003e(1):60.\u003c/li\u003e\n\u003cli\u003eChen Y, Hu J, Bu F, Zhang H, Fei K, Zhang P: \u003cstrong\u003eClinical characteristics of hyperprogressive disease in NSCLC after treatment with immune checkpoint inhibitor: a systematic review and meta-analysis\u003c/strong\u003e. \u003cem\u003eBMC cancer \u003c/em\u003e2020, \u003cstrong\u003e20\u003c/strong\u003e(1):707.\u003c/li\u003e\n\u003cli\u003e\u0026Aacute;lvarez-Velilla R, Guti\u0026eacute;rrez-Corbo MDC, Punz\u0026oacute;n C, P\u0026eacute;rez-Pertejo MY, Bala\u0026ntilde;a-Fouce R, Fresno M, Reguera RM: \u003cstrong\u003eA chronic bioluminescent model of experimental visceral leishmaniasis for accelerating drug discovery\u003c/strong\u003e. \u003cem\u003ePLoS neglected tropical diseases \u003c/em\u003e2019, \u003cstrong\u003e13\u003c/strong\u003e(2):e0007133.\u003c/li\u003e\n\u003cli\u003eLu L, Huang H, Zhou J, Ma W, Mackay S, Wang Z: \u003cstrong\u003eBRCA1 mRNA expression modifies the effect of T cell activation score on patient survival in breast cancer\u003c/strong\u003e. \u003cem\u003eBMC cancer \u003c/em\u003e2019, \u003cstrong\u003e19\u003c/strong\u003e(1):387.\u003c/li\u003e\n\u003cli\u003eZhao Y, Yu J, Zheng C, Zhou B: \u003cstrong\u003eEstablishment of a prognostic model for hypoxia-associated genes in OPSCC and revelation of intercellular crosstalk\u003c/strong\u003e. \u003cem\u003eFrontiers in immunology \u003c/em\u003e2024, \u003cstrong\u003e15\u003c/strong\u003e:1371365.\u003c/li\u003e\n\u003cli\u003eDe Zuani M, Xue H, Park JS, Dentro SC, Seferbekova Z, Tessier J, Curras-Alonso S, Hadjipanayis A, Athanasiadis EI, Gerstung M\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSingle-cell and spatial transcriptomics analysis of non-small cell lung cancer\u003c/strong\u003e. \u003cem\u003eNature communications \u003c/em\u003e2024, \u003cstrong\u003e15\u003c/strong\u003e(1):4388.\u003c/li\u003e\n\u003cli\u003eRitchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK: \u003cstrong\u003elimma powers differential expression analyses for RNA-sequencing and microarray studies\u003c/strong\u003e. \u003cem\u003eNucleic acids research \u003c/em\u003e2015, \u003cstrong\u003e43\u003c/strong\u003e(7):e47.\u003c/li\u003e\n\u003cli\u003eStuart T, Butler A, Hoffman P, Hafemeister C, Papalexi E, Mauck WM, 3rd, Hao Y, Stoeckius M, Smibert P, Satija R: \u003cstrong\u003eComprehensive Integration of Single-Cell Data\u003c/strong\u003e. \u003cem\u003eCell \u003c/em\u003e2019, \u003cstrong\u003e177\u003c/strong\u003e(7):1888-1902.e1821.\u003c/li\u003e\n\u003cli\u003eBunis DG, Andrews J, Fragiadakis GK, Burt TD, Sirota M: \u003cstrong\u003edittoSeq: universal user-friendly single-cell and bulk RNA sequencing visualization toolkit\u003c/strong\u003e. \u003cem\u003eBioinformatics (Oxford, England) \u003c/em\u003e2021, \u003cstrong\u003e36\u003c/strong\u003e(22-23):5535-5536.\u003c/li\u003e\n\u003cli\u003eAran D, Looney AP, Liu L, Wu E, Fong V, Hsu A, Chak S, Naikawadi RP, Wolters PJ, Abate AR\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eReference-based analysis of lung single-cell sequencing reveals a transitional profibrotic macrophage\u003c/strong\u003e. \u003cem\u003eNature immunology \u003c/em\u003e2019, \u003cstrong\u003e20\u003c/strong\u003e(2):163-172.\u003c/li\u003e\n\u003cli\u003eZhang X, Lan Y, Xu J, Quan F, Zhao E, Deng C, Luo T, Xu L, Liao G, Yan M\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eCellMarker: a manually curated resource of cell markers in human and mouse\u003c/strong\u003e. \u003cem\u003eNucleic acids research \u003c/em\u003e2019, \u003cstrong\u003e47\u003c/strong\u003e(D1):D721-d728.\u003c/li\u003e\n\u003cli\u003eKorsunsky I, Millard N, Fan J, Slowikowski K, Zhang F, Wei K, Baglaenko Y, Brenner M, Loh PR, Raychaudhuri S: \u003cstrong\u003eFast, sensitive and accurate integration of single-cell data with Harmony\u003c/strong\u003e. \u003cem\u003eNature methods \u003c/em\u003e2019, \u003cstrong\u003e16\u003c/strong\u003e(12):1289-1296.\u003c/li\u003e\n\u003cli\u003eJin S, Guerrero-Juarez CF, Zhang L, Chang I, Ramos R, Kuan CH, Myung P, Plikus MV, Nie Q: \u003cstrong\u003eInference and analysis of cell-cell communication using CellChat\u003c/strong\u003e. \u003cem\u003eNature communications \u003c/em\u003e2021, \u003cstrong\u003e12\u003c/strong\u003e(1):1088.\u003c/li\u003e\n\u003cli\u003eVan de Sande B, Flerin C, Davie K, De Waegeneer M, Hulselmans G, Aibar S, Seurinck R, Saelens W, Cannoodt R, Rouchon Q\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eA scalable SCENIC workflow for single-cell gene regulatory network analysis\u003c/strong\u003e. \u003cem\u003eNature protocols \u003c/em\u003e2020, \u003cstrong\u003e15\u003c/strong\u003e(7):2247-2276.\u003c/li\u003e\n\u003cli\u003eQiu X, Hill A, Packer J, Lin D, Ma YA, Trapnell C: \u003cstrong\u003eSingle-cell mRNA quantification and differential analysis with Census\u003c/strong\u003e. \u003cem\u003eNature methods \u003c/em\u003e2017, \u003cstrong\u003e14\u003c/strong\u003e(3):309-315.\u003c/li\u003e\n\u003cli\u003eHafemeister C, Satija R: \u003cstrong\u003eNormalization and variance stabilization of single-cell RNA-seq data using regularized negative binomial regression\u003c/strong\u003e. \u003cem\u003eGenome biology \u003c/em\u003e2019, \u003cstrong\u003e20\u003c/strong\u003e(1):296.\u003c/li\u003e\n\u003cli\u003eWu Y, Yang S, Ma J, Chen Z, Song G, Rao D, Cheng Y, Huang S, Liu Y, Jiang S\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSpatiotemporal Immune Landscape of Colorectal Cancer Liver Metastasis at Single-Cell Level\u003c/strong\u003e. \u003cem\u003eCancer discovery \u003c/em\u003e2022, \u003cstrong\u003e12\u003c/strong\u003e(1):134-153.\u003c/li\u003e\n\u003cli\u003eAndreatta M, Carmona SJ: \u003cstrong\u003eUCell: Robust and scalable single-cell gene signature scoring\u003c/strong\u003e. \u003cem\u003eComputational and structural biotechnology journal \u003c/em\u003e2021, \u003cstrong\u003e19\u003c/strong\u003e:3796-3798.\u003c/li\u003e\n\u003cli\u003eWolf FA, Angerer P, Theis FJ: \u003cstrong\u003eSCANPY: large-scale single-cell gene expression data analysis\u003c/strong\u003e. \u003cem\u003eGenome biology \u003c/em\u003e2018, \u003cstrong\u003e19\u003c/strong\u003e(1):15.\u003c/li\u003e\n\u003cli\u003ePeng G, Chi H, Gao X, Zhang J, Song G, Xie X, Su K, Song B, Yang J, Gu T\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eIdentification and validation of neurotrophic factor-related genes signature in HNSCC to predict survival and immune landscapes\u003c/strong\u003e. \u003cem\u003eFrontiers in genetics \u003c/em\u003e2022, \u003cstrong\u003e13\u003c/strong\u003e:1010044.\u003c/li\u003e\n\u003cli\u003eChen H, Yang W, Li Y, Ma L, Ji Z: \u003cstrong\u003eLeveraging a disulfidptosis-based signature to improve the survival and drug sensitivity of bladder cancer patients\u003c/strong\u003e. \u003cem\u003eFrontiers in immunology \u003c/em\u003e2023, \u003cstrong\u003e14\u003c/strong\u003e:1198878.\u003c/li\u003e\n\u003cli\u003eZhao J, Li G, Zhao G, Wang W, Shen Z, Yang Y, Huang Y, Ye L: \u003cstrong\u003ePrognostic signature of lipid metabolism associated LncRNAs predict prognosis and treatment of lung adenocarcinoma\u003c/strong\u003e. \u003cem\u003eFrontiers in oncology \u003c/em\u003e2022, \u003cstrong\u003e12\u003c/strong\u003e:986367.\u003c/li\u003e\n\u003cli\u003eChen B, Khodadoust MS, Liu CL, Newman AM, Alizadeh AA: \u003cstrong\u003eProfiling Tumor Infiltrating Immune Cells with CIBERSORT\u003c/strong\u003e. \u003cem\u003eMethods in molecular biology (Clifton, NJ) \u003c/em\u003e2018, \u003cstrong\u003e1711\u003c/strong\u003e:243-259.\u003c/li\u003e\n\u003cli\u003eMilke F, Wagner-Doebler I, Wienhausen G, Simon M: \u003cstrong\u003eSelection, drift and community interactions shape microbial biogeographic patterns in the Pacific Ocean\u003c/strong\u003e. \u003cem\u003eThe ISME journal \u003c/em\u003e2022, \u003cstrong\u003e16\u003c/strong\u003e(12):2653-2665.\u003c/li\u003e\n\u003cli\u003eKirkegaard RH, Dueholm MS, McIlroy SJ, Nierychlo M, Karst SM, Albertsen M, Nielsen PH: \u003cstrong\u003eGenomic insights into members of the candidate phylum Hyd24-12 common in mesophilic anaerobic digesters\u003c/strong\u003e. \u003cem\u003eThe ISME journal \u003c/em\u003e2016, \u003cstrong\u003e10\u003c/strong\u003e(10):2352-2364.\u003c/li\u003e\n\u003cli\u003eWang S, Chen Z, Lv H, Wang C, Wei H, Yu J: \u003cstrong\u003eLIPG is a novel prognostic biomarker and correlated with immune infiltrates in lung adenocarcinoma\u003c/strong\u003e. \u003cem\u003eJournal of clinical laboratory analysis \u003c/em\u003e2023, \u003cstrong\u003e37\u003c/strong\u003e(1):e24824.\u003c/li\u003e\n\u003cli\u003eBruno R, Poma AM, Panozzi M, Lenzini A, Elia G, Zirafa CC, Aprile V, Ambrogi MC, Baldini E, Lucchi M\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eEarly-Stage Non-Small Cell Lung Cancer: Prevalence of Actionable Alterations in a Monocentric Consecutive Cohort\u003c/strong\u003e. \u003cem\u003eCancers \u003c/em\u003e2024, \u003cstrong\u003e16\u003c/strong\u003e(7).\u003c/li\u003e\n\u003cli\u003eGao Z, Feng Y, Xu J, Liang J: \u003cstrong\u003eT-cell exhaustion in immune-mediated inflammatory diseases: New implications for immunotherapy\u003c/strong\u003e. \u003cem\u003eFrontiers in immunology \u003c/em\u003e2022, \u003cstrong\u003e13\u003c/strong\u003e:977394.\u003c/li\u003e\n\u003cli\u003eFribourg M, Anderson L, Fischman C, Cantarelli C, Perin L, La Manna G, Rahman A, Burrell BE, Heeger PS, Cravedi P: \u003cstrong\u003eT-cell exhaustion correlates with improved outcomes in kidney transplant recipients\u003c/strong\u003e. \u003cem\u003eKidney international \u003c/em\u003e2019, \u003cstrong\u003e96\u003c/strong\u003e(2):436-449.\u003c/li\u003e\n\u003cli\u003ePapaioannou E, Gonz\u0026aacute;lez-Molina MDP, Prieto-Mu\u0026ntilde;oz AM, G\u0026aacute;mez-Reche L, Gonz\u0026aacute;lez-Mart\u0026iacute;n A: \u003cstrong\u003eRegulation of Adaptive Tumor Immunity by Non-Coding RNAs\u003c/strong\u003e. \u003cem\u003eCancers \u003c/em\u003e2021, \u003cstrong\u003e13\u003c/strong\u003e(22).\u003c/li\u003e\n\u003cli\u003eLi J, Han T, Wang X, Wang Y, Yang R, Yang Q: \u003cstrong\u003eDevelopment of a CD8+\u003c/strong\u003e\u003cstrong\u003e\u0026thinsp;T cell associated signature for predicting the prognosis and immunological characteristics of gastric cancer by integrating single-cell and bulk RNA-sequencing\u003c/strong\u003e. \u003cem\u003eScientific reports \u003c/em\u003e2024, \u003cstrong\u003e14\u003c/strong\u003e(1):4524.\u003c/li\u003e\n\u003cli\u003eNakahara Y, Kouro T, Motoyama S, Miura M, Fujita K, Igarashi Y, Higashijima N, Matsuo N, Himuro H, Wei F\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eCirculating IL-6 and not its circulating signaling components sIL-6R and sgp130 demonstrate clinical significance in NSCLC patients treated with immune checkpoint inhibitors\u003c/strong\u003e. \u003cem\u003eFrontiers in cell and developmental biology \u003c/em\u003e2023, \u003cstrong\u003e11\u003c/strong\u003e:1324898.\u003c/li\u003e\n\u003cli\u003eLee HJ, Song KH, Oh SJ, Kim S, Cho E, Kim J, Park YG, Lee KM, Yee C, Song SH\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eTargeting TCTP sensitizes tumor to T cell-mediated therapy by reversing immune-refractory phenotypes\u003c/strong\u003e. \u003cem\u003eNature communications \u003c/em\u003e2022, \u003cstrong\u003e13\u003c/strong\u003e(1):2127.\u003c/li\u003e\n\u003cli\u003eSu S, Chen F, Lv X, Qi L, Ding Z, Ren W, Wei M, Liu Y, Yu L, Liu B\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003ePredictive value of peripheral blood biomarkers in patients with non-small-cell lung cancer responding to anti-PD-1-based treatment\u003c/strong\u003e. \u003cem\u003eCancer immunology, immunotherapy : CII \u003c/em\u003e2024, \u003cstrong\u003e73\u003c/strong\u003e(1):12.\u003c/li\u003e\n\u003cli\u003eOstroumov D, Fekete-Drimusz N, Saborowski M, K\u0026uuml;hnel F, Woller N: \u003cstrong\u003eCD4 and CD8 T lymphocyte interplay in controlling tumor growth\u003c/strong\u003e. \u003cem\u003eCellular and molecular life sciences : CMLS \u003c/em\u003e2018, \u003cstrong\u003e75\u003c/strong\u003e(4):689-713.\u003c/li\u003e\n\u003cli\u003eTang K, Zhang J, Cao H, Xiao G, Wang Z, Zhang X, Zhang N, Wu W, Zhang H, Wang Q\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eIdentification of CD73 as a Novel Biomarker Encompassing the Tumor Microenvironment, Prognosis, and Therapeutic Responses in Various Cancers\u003c/strong\u003e. \u003cem\u003eCancers \u003c/em\u003e2022, \u003cstrong\u003e14\u003c/strong\u003e(22).\u003c/li\u003e\n\u003cli\u003eKumar MA, Baba SK, Sadida HQ, Marzooqi SA, Jerobin J, Altemani FH, Algehainy N, Alanazi MA, Abou-Samra AB, Kumar R\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eExtracellular vesicles as tools and targets in therapy for diseases\u003c/strong\u003e. \u003cem\u003eSignal transduction and targeted therapy \u003c/em\u003e2024, \u003cstrong\u003e9\u003c/strong\u003e(1):27.\u003c/li\u003e\n\u003cli\u003eJung H, Paust S: \u003cstrong\u003eChemokines in the tumor microenvironment: implications for lung cancer and immunotherapy\u003c/strong\u003e. \u003cem\u003eFrontiers in immunology \u003c/em\u003e2024, \u003cstrong\u003e15\u003c/strong\u003e:1443366.\u003c/li\u003e\n\u003cli\u003eSathyanarayanan V, Neelapu SS: \u003cstrong\u003eCancer immunotherapy: Strategies for personalization and combinatorial approaches\u003c/strong\u003e. \u003cem\u003eMolecular oncology \u003c/em\u003e2015, \u003cstrong\u003e9\u003c/strong\u003e(10):2043-2053.\u003c/li\u003e\n\u003cli\u003eHagemann UB, Gunnarsson L, G\u0026eacute;raudie S, Scheffler U, Griep RA, Reiersen H, Duncan AR, Kiprijanov SM: \u003cstrong\u003eFully human antagonistic antibodies against CCR4 potently inhibit cell signaling and chemotaxis\u003c/strong\u003e. \u003cem\u003ePloS one \u003c/em\u003e2014, \u003cstrong\u003e9\u003c/strong\u003e(7):e103776.\u003c/li\u003e\n\u003cli\u003eFeng Y, He C, Liu C, Shao B, Wang D, Wu P: \u003cstrong\u003eExploring the Complexity and Promise of Tumor Immunotherapy in Drug Development\u003c/strong\u003e. \u003cem\u003eInternational journal of molecular sciences \u003c/em\u003e2024, \u003cstrong\u003e25\u003c/strong\u003e(12).\u003c/li\u003e\n\u003cli\u003eBarsch M, Sali\u0026eacute; H, Schlaak AE, Zhang Z, Hess M, Mayer LS, Tauber C, Otto-Mora P, Ohtani T, Nilsson T\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eT-cell exhaustion and residency dynamics inform clinical outcomes in hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eJournal of hepatology \u003c/em\u003e2022, \u003cstrong\u003e77\u003c/strong\u003e(2):397-409.\u003c/li\u003e\n\u003cli\u003eGanjoo S, Gupta P, Corbali HI, Nanez S, Riad TS, Duong LK, Barsoumian HB, Masrorpour F, Jiang H, Welsh JW\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eThe role of tumor metabolism in modulating T-Cell activity and in optimizing immunotherapy\u003c/strong\u003e. \u003cem\u003eFrontiers in immunology \u003c/em\u003e2023, \u003cstrong\u003e14\u003c/strong\u003e:1172931.\u003c/li\u003e\n\u003cli\u003eMotzer RJ, Martini JF, Mu XJ, Staehler M, George DJ, Valota O, Lin X, Pandha HS, Ching KA, Ravaud A: \u003cstrong\u003eMolecular characterization of renal cell carcinoma tumors from a phase III anti-angiogenic adjuvant therapy trial\u003c/strong\u003e. \u003cem\u003eNature communications \u003c/em\u003e2022, \u003cstrong\u003e13\u003c/strong\u003e(1):5959.\u003c/li\u003e\n\u003cli\u003eGrandin M, Mathot P, Devailly G, Bidet Y, Ghantous A, Favrot C, Gibert B, Gadot N, Puisieux I, Herceg Z\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eInhibition of DNA methylation promotes breast tumor sensitivity to netrin-1 interference\u003c/strong\u003e. \u003cem\u003eEMBO molecular medicine \u003c/em\u003e2016, \u003cstrong\u003e8\u003c/strong\u003e(8):863-877.\u003c/li\u003e\n\u003cli\u003eKim ST, Hong JY, Park SH, Park JO, Park YW, Park N, Lee H, Hong SH, Lee SJ, Song SW\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eFirst-in-human phase I trial of anti-hepatocyte growth factor antibody (YYB101) in refractory solid tumor patients\u003c/strong\u003e. \u003cem\u003eTherapeutic advances in medical oncology \u003c/em\u003e2020, \u003cstrong\u003e12\u003c/strong\u003e:1758835920926796.\u003c/li\u003e\n\u003cli\u003eLitak J, Czyzewski W, Szymoniuk M, Sakwa L, Pasierb B, Litak J, Hoffman Z, Kamieniak P, Rolinski J: \u003cstrong\u003eBiological and Clinical Aspects of Metastatic Spinal Tumors\u003c/strong\u003e. \u003cem\u003eCancers \u003c/em\u003e2022, \u003cstrong\u003e14\u003c/strong\u003e(19).\u003c/li\u003e\n\u003cli\u003eZhu X, Du J, Liu G: \u003cstrong\u003eThe comparison of multilineage differentiation of bone marrow and adipose-derived mesenchymal stem cells\u003c/strong\u003e. \u003cem\u003eClinical laboratory \u003c/em\u003e2012, \u003cstrong\u003e58\u003c/strong\u003e(9-10):897-903.\u003c/li\u003e\n\u003cli\u003eIshida S, Akiyama H, Umezawa Y, Okada K, Nogami A, Oshikawa G, Nagao T, Miura O: \u003cstrong\u003eMechanisms for mTORC1 activation and synergistic induction of apoptosis by ruxolitinib and BH3 mimetics or autophagy inhibitors in JAK2-V617F-expressing leukemic cells including newly established PVTL-2\u003c/strong\u003e. \u003cem\u003eOncotarget \u003c/em\u003e2018, \u003cstrong\u003e9\u003c/strong\u003e(42):26834-26851.\u003c/li\u003e\n\u003cli\u003eReolo MJY, Otsuka M, Seow JJW, Lee J, Lee YH, Nguyen PHD, Lim CJ, Wasser M, Chua C, Lim TKH\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eCD38 marks the exhausted CD8(+) tissue-resident memory T cells in hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eFrontiers in immunology \u003c/em\u003e2023, \u003cstrong\u003e14\u003c/strong\u003e:1182016.\u003c/li\u003e\n\u003cli\u003eMandal S, Ray B, Baniya Sharma S, Poulose J, Kasireddy V: \u003cstrong\u003eHyperprogression: A Unique Phenomenon of Progression of Existing Tumor Secondary to Immunotherapy\u003c/strong\u003e. \u003cem\u003eCureus \u003c/em\u003e2021, \u003cstrong\u003e13\u003c/strong\u003e(9):e17992.\u003c/li\u003e\n\u003cli\u003eZhang K, Chen H, Li F, Huang S, Chen F, Li Y: \u003cstrong\u003eBright future or blind alley? CAR-T cell therapy for solid tumors\u003c/strong\u003e. \u003cem\u003eFrontiers in immunology \u003c/em\u003e2023, \u003cstrong\u003e14\u003c/strong\u003e:1045024.\u003c/li\u003e\n\u003cli\u003eBedner E, Smolewski P, Amstad P, Darzynkiewicz Z: \u003cstrong\u003eActivation of caspases measured in situ by binding of fluorochrome-labeled inhibitors of caspases (FLICA): correlation with DNA fragmentation\u003c/strong\u003e. \u003cem\u003eExperimental cell research \u003c/em\u003e2000, \u003cstrong\u003e259\u003c/strong\u003e(1):308-313.\u003c/li\u003e\n\u003cli\u003ePunchihewa C, Inoue A, Hishiki A, Fujikawa Y, Connelly M, Evison B, Shao Y, Heath R, Kuraoka I, Rodrigues P\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eIdentification of small molecule proliferating cell nuclear antigen (PCNA) inhibitor that disrupts interactions with PIP-box proteins and inhibits DNA replication\u003c/strong\u003e. \u003cem\u003eThe Journal of biological chemistry \u003c/em\u003e2012, \u003cstrong\u003e287\u003c/strong\u003e(17):14289-14300.\u003c/li\u003e\n\u003cli\u003eTalwar H, McVicker B, Tobi M: \u003cstrong\u003ep38\u0026gamma; Activation and BGP (Biliary Glycoprotein) Induction in Primates at Risk for Inflammatory Bowel Disease and Colorectal Cancer-A Comparative Study with Humans\u003c/strong\u003e. \u003cem\u003eVaccines \u003c/em\u003e2020, \u003cstrong\u003e8\u003c/strong\u003e(4).\u003c/li\u003e\n\u003cli\u003eZhang X, Goldstein DY, Khader SN: \u003cstrong\u003eEducational Case: Non-Small Cell Lung Cancer: Pathologic Diagnosis and Molecular Understanding\u003c/strong\u003e. \u003cem\u003eAcademic pathology \u003c/em\u003e2019, \u003cstrong\u003e6\u003c/strong\u003e:2374289519881951.\u003c/li\u003e\n\u003cli\u003eDengler MA, Teh CE, Thijssen R, Gangoda L, Lan P, Herold MJ, Gray DH, Kelly GL, Roberts AW, Adams JM: \u003cstrong\u003ePotent efficacy of MCL-1 inhibitor-based therapies in preclinical models of mantle cell lymphoma\u003c/strong\u003e. \u003cem\u003eOncogene \u003c/em\u003e2020, \u003cstrong\u003e39\u003c/strong\u003e(9):2009-2023.\u003c/li\u003e\n\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":"NSCLC, T cell exhaustion, scRNA-seq, Spatial transcriptomics and Prognostic models","lastPublishedDoi":"10.21203/rs.3.rs-7434355/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7434355/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eNon-small cell lung cancer (NSCLC) represents 85% of lung cancers and remains a leading cause of cancer death. Immunotherapy advancements have improved treatment, but many patients develop resistance due to T cell exhaustion. Understanding this mechanism, aided by single-cell RNA sequencing, is vital for creating personalized therapies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eSingle-cell RNA sequencing (scRNA-seq) and bulk RNA-seq data from NSCLC patients and normal tissues were collected from multiple databases. Batch effects were corrected, and scRNA-seq data were processed using Seurat for dimensionality reduction and clustering. Exhausted T cell subpopulations were identified, and transcriptional and spatial analyses were conducted using SCENIC and pseudotime analysis. Additionally, a 20-T-ExhauRs prognostic model was developed using machine learning algorithms, and immune infiltration and drug sensitivity analyses were performed. Statistical analyses were conducted using R and Python software.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e The study identified exhausted T cell subpopulations in NSCLC using dimensionality reduction and clustering, revealing 25 subpopulations and significant differences between normal and NSCLC groups. Pseudotime and transcription factor analysis showed the evolution of exhausted T cell subpopulations. Spatial transcriptomics and metabolic pathway enrichment revealed heterogeneity in the tumor microenvironment. The 20-T-ExhauRs prognostic model was developed using machine learning and demonstrated strong survival prediction accuracy. Immune infiltration analysis revealed weaker immune responses in high-risk groups, while drug sensitivity analysis indicated reduced effectiveness of certain chemotherapies. The study offers insights into immune regulation, tumor progression, and therapeutic strategies for NSCLC.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion: \u003c/strong\u003eThis study identified exhausted T cell subpopulations in NSCLC, revealing their roles in tumor progression. The 20-T-ExhauRs model accurately predicted survival outcomes. Spatial transcriptomics and immune infiltration analyses highlighted tumor heterogeneity, suggesting potential therapeutic strategies to improve NSCLC treatment and patient prognosis.\u003c/p\u003e","manuscriptTitle":"Dissecting T Cell Exhaustion in Non-Small Cell Lung Cancer: Single-Cell and Spatial Transcriptomics Reveal Prognostic Signatures and Therapeutic Implications","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-06 09:18:47","doi":"10.21203/rs.3.rs-7434355/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":"cc6b5a2a-fba1-4242-a5b3-4d788ee55e3a","owner":[],"postedDate":"October 6th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-11-12T09:25:00+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-06 09:18:47","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7434355","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7434355","identity":"rs-7434355","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}