Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Exploring hypoxia-related genes as prognostic indicators in lung adenocarcinoma

  • Bandar Alghamdi,

    Roles Conceptualization, Data curation, Formal analysis, Methodology, Visualization, Writing – original draft

    Affiliation King Fahad Army Forces Hospital, Cardiology and Cardiac Surgery Center, Jeddah, Saudi Arabia

    ⨯
  • Sonia Rocha

    Roles Funding acquisition, Supervision, Writing – review & editing

    srocha@liverpool.ac.uk

    Affiliation Department of Biochemistry, Cell and Systems Biology, Institute of Systems Molecular and Integrative Biology, University of Liverpool, Liverpool, United Kingdom

    ⨯

Abstract

Background

Lung adenocarcinoma (LUAD) is a leading cause of cancer-related mortality, with hypoxia contributing to tumor progression and treatment resistance. Identifying hypoxia-related biomarkers could enhance prognosis and therapeutic strategies for LUAD.

Methods

This study aimed to identify hypoxia-related differentially expressed genes (HRDEGs) in lung adenocarcinoma (LUAD) through differential expression analysis. Functional analysis and protein-protein interaction (PPI) network construction were performed to explore the biological roles and interactions of these genes. Kaplan-Meier survival analysis, univariate Cox regression, and Lasso regression were used to identify key genes associated with survival. Multivariate Cox regression was then conducted to assess independent prognostic factors.

Results

This analysis revealed 283 upregulated HRDEGs and 322 downregulated HRDEGs in LUAD. Functional enrichment analysis indicated that the upregulated genes were primarily involved in cancer-related and cellular signaling pathways, while downregulated genes were associated with immunity-related pathways. We further identified 201 common upregulated hub genes (including MMP9, CDH1, HSP90AB1, SOX2, CDKN2A, SPP1, EZH2) and 224 common downregulated hub genes (such as IL6, TNF, IL1B, JUN, CCL2, TLR4, FOS, PTGS2). Kaplan-Meier survival analysis, univariate Cox regression, and Lasso regression led to the identification of 17 key genes (ADRB2, ALDH2, CAT, CCNE1, MAP3K8, DSG2, EIF6, ABCB1, PIK3R1, RAD51, SFTPD, SOD3, CLEC3B, ADAM12, EXO1, FBLN5, and IGF2BP3) associated with patient survival. Finally, multivariate Cox regression analysis identified DSG2, EIF6, and EXO1 as independent prognostic factors for LUAD, highlighting their potential as biomarkers for prognosis and therapeutic targets in lung cancer.

Conclusion

In conclusion, DSG2, EIF6, and EXO1 were identified as key hypoxia-related genes in lung adenocarcinoma. These genes were found to be independent prognostic factors, highlighting their potential as biomarkers for predicting patient survival and guiding future therapeutic approaches.

Citation: Alghamdi B, Rocha S (2026) Exploring hypoxia-related genes as prognostic indicators in lung adenocarcinoma. PLoS One 21(6): e0349820. https://doi.org/10.1371/journal.pone.0349820

Editor: Hongtao Bi, Northwest Institute of Plateau Biology Chinese Academy of Sciences, CHINA

Received: September 17, 2025; Accepted: May 4, 2026; Published: June 3, 2026

Copyright: © 2026 Alghamdi, Rocha. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are publicly available. The transcriptomic and genomic datasets analyzed in this study were retrieved from publicly accessible repositories, including GEO (Gene Expression Omnibus) and TCGA (The Cancer Genome Atlas). Specific accession numbers and dataset identifiers are provided within the manuscript. Accession numbers and links are: GSE18842; GSE13213; https://www.cbioportal.org/study/summary?id=luad_tcga. No new data were generated.

Funding: This study was supported by the Wellcome Trust to SR (206293/Z/17/Z). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Lung Adenocarcinoma and Hypoxia Lung cancer remains one of the most common and deadly cancers worldwide, with non-small cell lung cancer (NSCLC) being the dominant histological subtype [1]. Lung adenocarcinoma (LUAD) is the most prevalent histological subtype of NSCLC and represents a major clinical and biological focus within this disease category due to its distinct molecular and genetic characteristics. Although advances in molecular profiling and targeted therapies have improved the management of NSCLC, the prognosis for patients with LUAD remains poor, particularly in advanced stages [2]. The overall five-year survival rate for LUAD patients is low, making it essential to identify novel biomarkers for early diagnosis, prognosis, and therapeutic targeting [3]. One of the critical microenvironments influencing cancer progression, metastasis, and resistance to therapy is tumor hypoxia. Hypoxia refers to the state of low oxygen availability within the tumor, often resulting from an imbalance between the rapid growth of tumor cells and the insufficient development of blood vessels to supply oxygen [4]. Hypoxia is a hallmark feature of many solid tumors, including lung adenocarcinoma, and it is associated with aggressive tumor behavior, poor prognosis, and resistance to conventional therapies [5]. The development of therapeutic strategies targeting hypoxic tumor cells remains an area of active research.

Hypoxia plays a crucial role in driving various aspects of tumor biology, such as tumor angiogenesis, metabolic reprogramming, and epithelial-to-mesenchymal transition (EMT) [6]. It activates a range of adaptive cellular responses through hypoxia-inducible factors (HIFs), primarily HIF-1α and HIF-2α [7]. These transcription factors regulate the expression of genes that mediate critical processes such as angiogenesis (formation of new blood vessels), glycolysis, metastasis, and drug resistance [8,9]. Tumor cells exposed to hypoxic conditions undergo metabolic shifts that favor anaerobic glycolysis, which allows them to survive and proliferate despite low oxygen availability. The HIF signaling pathway is essential for the regulation of genes involved in cellular survival, metabolism, and tumor progression [8,10]. Among the most well-known hypoxia-regulated genes are vascular endothelial growth factor (VEGF), glucose transporter proteins (e.g., GLUT1), and matrix metalloproteinases (e.g., MMPs) [11]. These genes promote tumor growth and metastasis by facilitating the adaptation of tumor cells to low-oxygen conditions [12–14]. Additionally, hypoxia is implicated in the regulation of immune evasion and resistance to chemotherapies and radiation therapy, further complicating treatment strategies [15]. Therefore, the identification of key hypoxia-related genes and their role in LUAD progression and prognosis is crucial for improving the clinical management of lung adenocarcinoma.

The link between hypoxia and poor prognosis in LUAD has been well established through various studies. Tumors with hypoxic regions have been shown to have a higher potential for metastasis, increased tumor aggressiveness, and reduced sensitivity to treatment [16]. Patients with high expression levels of hypoxia-related genes often exhibit shorter overall survival and progression-free survival compared to those with low expression levels of these genes in many cancers [17]. Several studies have sought to identify specific hypoxia-related biomarkers that could serve as predictors of patient outcomes in LUAD [18–20]. Li et al. analyzed immune–hypoxia-related immune genes and developed a 7-gene prognostic signature; while informative, their study focused narrowly on genes intersecting immune and hypoxia pathways, without comprehensive assessment of hub genes or functional interactions [18]. Xiong et al. identified 15 hypoxia-related genes to classify hypoxia clusters and constructed a ceRNA-based prognostic network; however, their analysis prioritized clustering and non-coding RNA networks, with limited evaluation of independent prognostic value of core genes [19]. The Malawi Medical Journal study established a 9-lncRNA hypoxia signature to predict survival, but focused exclusively on non-coding RNAs and lacked broader gene-level validation across multiple cohorts [20]. Collectively, these studies provided valuable insights into hypoxia-driven mechanisms, but none performed a genome-wide identification of hypoxia-related genes, systematically integrated PPI networks, and validated independent prognostic genes across multiple datasets. In contrast, the present study performs a comprehensive, multilayered analysis of hypoxia-related differentially expressed genes, integrating functional enrichment, protein–protein interaction networks, and survival modeling to identify independent prognostic hub genes. This approach not only captures broader hypoxia-induced changes in LUAD but also provides mechanistic insights into key pathways and biological processes, thereby extending the current understanding of hypoxia-driven tumor progression and prognosis.

However, the heterogeneity of LUAD, coupled with the complexity of hypoxia-induced changes in gene expression, presents challenges in pinpointing specific prognostic genes. Additionally, the tumor microenvironment in LUAD is influenced by various factors, such as genetic mutations, immune responses, and stromal interactions, which may confound the impact of hypoxia on survival [21,22]. While some studies focused on a small number of hypoxia-induced genes [19], other studies were interested in immune phenotypes [18,19] and non-coding RNAs [19,20] as potential biomarkers of the disease. These studies, using robust statistics and analysis pipelines produced valid and potential biomarkers in these subareas of hypoxia induced changes, with varying prognostic value. However, they all lacked validation in independent cohort of data. Furthermore, by using prior selection of given criteria such small number of hypoxia-induced genes, focusing mostly on non-coding RNAs, these studies increased biases and also might have lost critical information in other areas of the data available.

In this context, a comprehensive analysis of hypoxia-related genes and their potential prognostic value in LUAD is thus still missing. Identifying genes that are consistently associated with poor prognosis, regardless of other clinical variables, will allow for the development of better prognostic models. These models could guide personalized treatment strategies, such as targeting hypoxic regions of the tumor with hypoxia-activated pro-drugs or improving the efficacy of therapies through the modulation of the HIF pathway.

This study aims to systematically investigate hypoxia-related genes in LUAD and evaluate their potential prognostic significance. We used a combination of computational and statistical approaches to analyze large-scale genomic datasets, while also validating the prognostic value of key findings in an independent cohort of patients. By identifying key hypoxia-associated genes, this work seeks to find independent prognostic biomarkers and provide insights into hypoxia-driven mechanisms underlying disease progression in LUAD Our iterative and multilayered analysis lead to the identification of three genes that held across all datasets analyzed and had predictive prognostic value in LUAD.

Methods

Data downloading and processing

The analysis utilized two distinct datasets: the TCGA LUAD cohort and GSE18842. We downloaded the TCGA LUAD cohort form the cBioPortal [23]. The TCGA LUAD cohort comprised 517 tumor samples and 59 control samples, representing lung adenocarcinoma (LUAD) patients and normal controls, respectively (https://www.cbioportal.org/study/summary?id=luad_tcga). Clinical data (https://www.cbioportal.org/study/summary?id=luad_tcga) from this cohort were incorporated for Kaplan-Meier survival analysis, univariate regression analysis, and multivariate regression analysis to investigate the relationship between gene expression and survival outcomes. For the GSE18842 dataset [24], which contains 46 tumor samples and 45 control samples. data normalization was performed using a log2 transformation. To account for multiple probes corresponding to individual genes, the data was averaged, resulting in a single expression value for each gene. The GSE18842 dataset is publicly available and can be accessed at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE18842.

Identification of differentially expressed gene (DEGs) and hypoxia related differentially expressed gene (HRDEFs)

In this study, we utilized multiple datasets to identify differentially expressed genes (DEGs) associated with tumorigenesis. To identify DEGs between tumor and normal samples, we employed the LIMMA package [25], applying a stringent cutoff of |log2FC| > 0.585 and an adjusted p-value < 0.05. Following DEG identification, we specifically targeted hypoxia-related differentially expressed genes (HRDEGs), which were selected through a Venn diagram approach. This method allowed us to cross-reference the LUAD DEGs and further refine the set of genes related to hypoxia, thus providing a focused analysis of the genes potentially involved in the tumor microenvironment’s hypoxic conditions.

Collection of hypoxia related genes and identification of hypoxia related differentially expressed genes (HRDEGs)

To curate a comprehensive list of hypoxia-related genes, we integrated data from several established resources. First, we retrieved hypoxia-related genes from Gene Set Enrichment Analysis (GSEA) (https://www.gsea-msigdb.org/gsea/index.jsp) [26], focusing on the following gene sets: GO BP: hypoxia inducible factor 1 alpha signaling pathway, GO BP: Intrinsic apoptotic signaling pathway in response to hypoxia, GO BP: Negative regulation of cellular response to hypoxia, GO BP: Negative regulation of hypoxia induced intrinsic apoptotic signaling pathway, GO BP: Regulation of cellular response to hypoxia, hallmark hypoxia, and Reactome: Regulation of gene expression by hypoxia inducible factor. We then performed a search on OMIM (https://www.ncbi.nlm.nih.gov/omim/?term=HYPOXIA) using the term “HYPOXIA” to identify genes associated with hypoxia-related genetic disorders. Additionally, we retrieved genes from the hsa04066:HIF-1 signaling pathway in the KEGG database (https://www.genome.jp/kegg/). Finally, we obtained genes from GeneCards (https://www.genecards.org/), selecting those with a score greater than or equal to 1. After compiling the data from these sources, we performed deduplication to ensure that only unique genes were included. This resulted in a final list of 3292 unique genes, which are tabulated in Supplementary Table S1 in S1 File.

Functional enrichment analysis of DEGs

We used DAVID v6.8 (https://david.ncifcrf.gov/) [27] to perform functional annotation of the identified genes. This platform allowed us to categorize the genes into various biological contexts, such as significantly enriched KEGG pathways, biological processes, cellular components, and molecular functions. These functional analyses provided valuable insights into the roles and interactions of the genes in key cellular mechanisms. To assess the statistical significance of the enrichment, we applied a false discovery rate (FDR) threshold of < 0.05, calculated using the Benjamini–Hochberg method [28], ensuring that only robust and biologically relevant findings were included.

Exploration of hub HRDEGs

We constructed Protein-Protein Interaction (PPI) networks for the differentially expressed genes (DEGs) using STRING v11.0 (https://string-db.org/) [29]. Both upregulated and downregulated genes were entered into the STRING search module to generate the interaction networks. To identify key hub genes within these networks, we utilized the cytoHubba plugin in Cytoscape (https://cytoscape.org/) [30], which ranks genes based on their interaction degree. A medium interaction score threshold of ≥0.40 was applied to build the PPI network, and genes with an interaction degree of ≥10 were selected for further analysis. This approach enabled us to focus on the most significant hub genes, highlighting those with strong interactions relevant to lung cancer.

Exploration of survival associated hub HRDEGs

We retrieved clinical data from TCGA LUAD to perform an analysis of overall survival (OS) in relation to gene expression. Patients were categorized into two groups based on their gene expression levels, either above or below the median, and survival rates between these groups were compared. To further explore survival differences, patients were also stratified into risk groups based on their gene expression profiles. The log-rank test and Kaplan-Meier survival curves were utilized to evaluate and visualize differences in survival outcomes. For a more detailed survival analysis, we used the R package “survival” [31], performing both univariate and multivariate Cox regression analyses to assess the independent effects of genes on survival. Additionally, we applied LASSO-Cox regression via the “glmnet” package [32], incorporating 10-fold cross-validation to optimize the selection of penalty parameters and minimize the risk of overfitting. This comprehensive approach allowed us to identify significant survival-associated genes while controlling for potential biases and overfitting, ensuring robust and reliable findings.

Validation of prognostic signature in an independent cohort

To validate the prognostic relevance of the identified three-gene signature, we utilized the independent LUAD dataset GSE13213 via the SurvExpress platform [33]. SurvExpress is an online tool that allows assessment of gene expression-based prognostic models across independent cohorts. The multivariate Cox regression model included only the expression levels of these three genes. The expression levels of gene signature were used to calculate a risk score for each patient, and patients were stratified into high-risk and low-risk groups based on the median risk score. Survival differences between the groups were evaluated using Kaplan-Meier survival analysis and the Log-Rank test, and hazard ratios (HRs) with 95% confidence intervals were computed to assess prognostic significance. Additionally, expression differences of gene signature between the risk groups were compared to confirm consistency with the primary cohort findings.

Statistical analysis and overall methodology

All statistical analyses were performed using R software (version 4.0.1). Differentially expressed genes (DEGs) were identified using a cutoff threshold of |log2 fold change (log2FC)| > 0.585 and an adjusted p-value < 0.05. Multiple testing correction was conducted using the Benjamini–Hochberg false discovery rate (FDR) method, with an FDR threshold of < 0.05 applied to control for false positives [28]. A p-value threshold of 0.05 was used for the survival analysis of HRDEGs, applied to Kaplan–Meier method, log-rank test, and Cox regression analysis. The overall methodology provided in the Fig 1.

thumbnail
Download:
Fig 1. The overall methodology of the study.

https://doi.org/10.1371/journal.pone.0349820.g001

Results

Identification of hypoxia related differentially expressed genes (HRDEGs) in Lung adenocarcinoma

We explored differentially expressed genes (DEGs) in the TCGA LUAD cohort (Supplementary Table S2 in S1 File) and identified additional DEGs from the GSE18842 dataset (Supplementary Table S3 in S1 File). Volcano plots depicting the results of both analyses are shown in Fig 2A and Fig 2B, respectively. These plots provide a visual representation of the relationship between fold changes (logFC) and adjusted p-values, helping to identify genes with significant differential expression in lung adenocarcinoma (LUAD). Next, we compared the identified DEGs with hypoxia-related genes to determine the hypoxia-related differentially expressed genes (HRDEGs). This comparison revealed a total of 283 upregulated HRDEGs and 322 downregulated HRDEGs in LUAD, which are displayed in Fig 2C-2D and detailed in Supplementary Table S4 in S1 File. These findings indicate that specific hypoxia-related genes are significantly altered in LUAD, providing potential insights into the molecular mechanisms underlying the disease. The top ten upregulated and downregulated HRDEGs are listed in Table 1. Among the top upregulated HRDEGs are CYP24A1, CA9, MMP13, SPP1, COL10A1, MMP1, EEF1A2, GREM1, PITX1, and MMP12. These genes are associated with various biological processes such as cell metabolism, matrix remodeling, and tumor progression. On the other hand, the top downregulated HRDEGs include SFTPC, SLC6A4, AGER, FABP4, SCGB1A1, VEGFD, HBB, SFTPA2, AQP4, and CLEC3B, which are involved in processes like lung surfactant production, cell signaling, and vasculature maintenance.

thumbnail
Download:
Table 1. The top ten upregulated and downregulated HRDEGs in TCGA and GSE18842.

https://doi.org/10.1371/journal.pone.0349820.t001

thumbnail
Download:
Fig 2. Volcano plots and Venn diagrams depicting DEGs and HRDEGs in LUAD.

A. Volcano plot of DEGs in the TCGA LUAD cohort. B. Volcano plot of DEGs in the GSE18842 dataset, illustrating differential gene expression in LUAD. C. Venn diagram showing the overlap of the upregulated hypoxia related differentially expressed genes (HRDEGs) in LUAD. D. Venn diagram illustrating the overlap of downregulated HRDEGs in LUAD.TCGA UP: TCGA-LUAD upregulated; TCGA DOWN: TCGA-LUAD downregulated; HRG: hypoxia-related genes.

https://doi.org/10.1371/journal.pone.0349820.g002

HRDEGs are associated with cellular processes such as cell cycle and metabolism

Fig 3 illustrates the pathways significantly enriched by the upregulated genes, with a false discovery rate (FDR) < 0.05. These pathways highlight key biological processes and molecular mechanisms potentially associated with tumorigenesis (Supplementary Table S5 in S1 File). Notably, the cell cycle (hsa04110), biosynthesis of amino acids (hsa01230), and carbon metabolism (hsa01200) are central to cellular growth and energy production. The p53 signaling pathway (hsa04115) and cellular senescence (hsa04218) are also implicated, reflecting the regulation of cell death and aging processes. Other enriched pathways, such as glycolysis/gluconeogenesis (hsa00010) and HIF-1 signaling (hsa04066), suggest alterations in metabolic pathways and hypoxic responses often observed in cancer. Additionally, pathways related to cancer progression, such as central carbon metabolism in cancer (hsa05230), pathways in cancer (hsa05200), and microRNAs in cancer (hsa05206), are also prominently featured, highlighting the complex molecular interactions in cancer biology. Other enriched pathways include DNA replication (hsa03030), mismatch repair (hsa03430), and glutathione metabolism (hsa00480), which are critical for genome stability and oxidative stress responses. The statistical analysis of upregulated pathways is tabulated in Supplementary Table S5 in S1 File. These results emphasize the multifaceted nature of the upregulated genes and their involvement in various cellular functions relevant to cancer development.

thumbnail
Download:
Fig 3. Significantly enriched pathways associated with the upregulated genes (FDR < 0.05).

DEG upregulated were analyzed for functional pathways using DAVID.

https://doi.org/10.1371/journal.pone.0349820.g003

The top enriched pathway, the Cell Cycle (Fig 4), with a highly significant p-value of 4.51E-06. This pathway is critical for regulating cell division and proliferation. The genes enriched within this pathway, including PCNA, MCM7, CDKN2A, PRKDC, HDAC1, CDC6, and others such as CDK4, CHEK1, and E2F3, play pivotal roles in cell cycle aberration. These genes are integral to maintaining proper cell cycle regulation, and their upregulation suggests a potential disruption in normal cell division processes, which is commonly associated with tumorigenesis. The enrichment of these genes further underscores the importance of the cell cycle in hypoxic lung cancer biology.

thumbnail
Download:
Fig 4. KEGG pathway illustrating the cell cycle pathway from DAVID, with upregulated genes marked by red stars.

https://doi.org/10.1371/journal.pone.0349820.g004

Similar to the Cell Cycle pathway, another significantly enriched pathway is the HIF-1 signaling pathway (hsa04066), with genes such as LDHA, EGLN3, SLC2A1, EIF4EBP1, PGK1, TIMP1, ENO1, ALDOA, GAPDH, PFKP, and PDK1 being upregulated (Fig 5). The upregulation of these genes suggests a critical role of hypoxia in lung cancer, as these genes are involved in metabolic adaptation to low oxygen conditions [34], many of which are known HIF-dependent targets [12]. This highlights the importance of hypoxia-induced signaling in promoting tumor survival and progression through altered cellular metabolism.

thumbnail
Download:
Fig 5. KEGG pathway illustrating the HIF-1 signaling pathway from DAVID, with upregulated genes marked by red stars.

Upregulated genes, such as LDHA, EGLN3, SLC2A1, EIF4EBP1, PGK1, TIMP1, ENO1, ALDOA, GAPDH, PFKP, and PDK1, are shown with red stars, indicating their involvement in metabolic reprogramming and cellular adaptation to low oxygen conditions.

https://doi.org/10.1371/journal.pone.0349820.g005

In addition to the upregulated genes, we observed that the downregulated genes are predominantly associated with immune-related pathways and signaling processes. As illustrated in Fig 6, several immune-related pathways are significantly affected, indicating a potential suppression of immune responses in lung cancer. Key pathways include the TNF signaling pathway (hsa04668), Cytokine-cytokine receptor interaction (hsa04060), and IL-17 signaling pathway (hsa04657), all of which play essential roles in immune activation and inflammation. Other important pathways such as Toll-like receptor signaling (hsa04620), JAK-STAT signaling (hsa04630), and Natural killer cell-mediated cytotoxicity (hsa04650) are also downregulated, suggesting that the immune system’s ability to recognize and eliminate cancer cells may be impaired. The statistical analysis of upregulated pathways is tabulated in Supplementary Table S6 in S1 File. These findings suggest a potential immune evasion mechanism in lung cancer, where the downregulation of key immune pathways may allow tumors to escape immune surveillance and progress unchecked.

thumbnail
Download:
Fig 6. Immune-associated pathways enriched in the downregulated genes.

This figure illustrates the major immune-related pathways that are significantly downregulated in lung cancer.

https://doi.org/10.1371/journal.pone.0349820.g006

Explorations of hub HRDEGs

Next, we conducted a Protein-Protein Interaction (PPI) analysis by inputting all the hypoxia-related differentially expressed genes (DEGs) into the STRING database. The primary PPI network consisted of 600 nodes and 9,281 edges, with an average node degree of 30.9 (Fig 7). The PPI enrichment p-value was found to be < 1.0e-16, indicating a highly significant interaction among the hypoxia-related genes. Based on a degree of interaction greater than 10, we identified the hub genes within the network. After identifying the hub genes based on the degree of interaction, we found that the top hub genes include GAPDH, IL6, TNF, IL1B, MMP9, JUN, CCL2, TLR4, FOS, PTGS2, PECAM1, CDH1, HSP90AB1, PTPRC, ICAM1, FGF2, PPARG, ITGAM, CXCL12, and KDR. The list of all interacted genes and their degree of interaction provided in the Supplementary Table S7 in S1 File. These genes play pivotal roles in key biological processes such as inflammation, immune response, cell signaling, and tumor progression. Fig 8A illustrates the interaction network of the top 50 hub genes an, highlighting their interconnections and potential involvement in the molecular pathways that drive lung cancer progression. We further identified 201 common upregulated hub genes and 224 common downregulated hub genes, which were shared across the datasets. These genes represent key molecular players in the tumorigenic processes of lung cancer, with their dysregulation potentially contributing to cancer progression. The results of this analysis are presented in Fig 8B and are detailed in Supplementary Table S7 in S1 File.

thumbnail
Download:
Fig 7. Primary Protein-Protein Interaction (PPI) network of hypoxia-related differentially expressed genes (DEGs) generated using STRING.

The network consists of 600 nodes and 9,281 edges, with an average node degree of 30.9.

https://doi.org/10.1371/journal.pone.0349820.g007

thumbnail
Download:
Fig 8. Protein-Protein Interaction (PPI) Network of Hub Genes.

A. Interaction network of the top 50 hub genes identified from the PPI analysis. B. Common hub genes shared between the upregulated and downregulated gene sets. Yellow node represents upregulated genes and light blue node represents downregulated genes.

https://doi.org/10.1371/journal.pone.0349820.g008

The top upregulated hub genes identified in this analysis include MMP9, CDH1, HSP90AB1, SOX2, CDKN2A, SPP1, EZH2, TIMP1, CXCL10, COL1A1, HDAC1, BRCA1, and CDK4. These genes are involved in critical processes such as tumor invasion, immune modulation, cell cycle regulation, and epigenetic changes. The top downregulated hub genes identified in this analysis include IL6, TNF, IL1B, JUN, CCL2, TLR4, FOS, PTGS2, PECAM1, PTPRC, ICAM1, FGF2, PPARG, ITGAM, CXCL12, and KDR. These genes are primarily involved in immune response, inflammation, and cellular signaling pathways.

Exploration of independent prognostic HRDEGs

We first conducted Kaplan-Meier survival analysis on all hub genes and identified 174 significant genes that were associated with survival (Supplementary Table S8 in S1 File). The following seven genes (Identified by Univariate regression and LASO)—ADAM12 (HR = 1.39, P = 0.02), CCNE1 (HR = 1.43, P = 0.01), DSG2 (HR = 1.58, P = 0.002), EIF6(HR = 1.49, P = 0.008), EXO1 (HR = 1.54, P = 0.003), IGF2 BP3 (HR = 1.6, P = 0.001), and RAD51 (HR = 1.37, P = 0.03)—are identified as upregulated hub genes in lung cancer (Fig 9A).

thumbnail
Download:
Fig 9. Kaplan-Meier Survival Analysis of Upregulated and Downregulated Hub Genes in Lung Cancer.

A. Kaplan-Meier survival analysis of the group of upregulated hub genes, showing their significant association with survival in lung cancer patients. (B) Kaplan-Meier survival analysis of the group of downregulated hub genes, demonstrating their significant impact on survival outcomes.

https://doi.org/10.1371/journal.pone.0349820.g009

The upregulated group, categorized based on median expression levels, is associated with worse survival outcomes. Increased expression of these genes is correlated with poorer prognosis, suggesting that they may play a critical role in tumor progression and could serve as potential biomarkers for identifying high-risk patients with lung cancer. The following ten genes ((Identified by Univariate regression and LASO))—ABCB1(HR = 0.71, P = 0.02), ADRB2(HR = 0.67, P = 0.008), ALDH2(HR = 0.66, P = 0.006), CAT(HR = 0.73, P = 0.03), CLEC3B(HR = 0.67, P = 0.006), FBLN5(HR = 0.64, P = 0.003), MAP3K8(HR = 0.67, P = 0.006), PIK3R1(HR = 0.64, P = 0.003), SFTPD(HR = 0.68, P = 0.01), and SOD3(HR = 0.75, P = 0.04)—are identified as downregulated hub genes in lung cancer.

To refine this list, we performed univariate regression analysis, which revealed 106 genes with statistically significant associations (Supplementary Table S9 in S1 File). Subsequently, we applied Lasso regression to the consistently upregulated and downregulated genes to identify key genes with non-zero coefficients. This analysis led to the identification of 17 key genes: ADRB2, ALDH2, CAT, CCNE1, MAP3K8, DSG2, EIF6, ABCB1, PIK3R1, RAD51, SFTPD, SOD3, CLEC3B, ADAM12, EXO1, FBLN5, and IGF2BP3. The univariate regression analysis and LASSO regression analysis depicted in the Figs 10 and 11, respectively.

thumbnail
Download:
Fig 10. The results of univariate Cox regression analysis for hub genes in lung cancer patients.

Each gene’s hazard ratio (HR) with 95% confidence interval (CI) is displayed, indicating the association between gene expression and patient survival.

https://doi.org/10.1371/journal.pone.0349820.g010

thumbnail
Download:
Fig 11. Lasso regression analysis of key genes.

Identification of key genes through Lasso regression. (A) Lasso coefficients for each gene selected through Lasso regression. The plot shows the evolution of the coefficients for each gene as the regularization penalty (lambda) increases. (B) Log(lambda) vs. coefficients plot demonstrating the relationship between the log-transformed lambda values and the corresponding coefficients.

https://doi.org/10.1371/journal.pone.0349820.g011

Furthermore, a multivariate Cox regression analysis was conducted to identify independent prognostic factors. The results indicated that DSG2, EIF6, and EXO1 emerged as independent prognostic factors (Fig 12). The multivariate model demonstrated a significant fit, with the following statistics: Concordance = 0.657 (SE = 0.024), Likelihood ratio test = 50.39 on 17 degrees of freedom (df), p = 4e-05, Wald test = 48.47 on 17 df (p = 7e-05), and Score (logrank) test = 48.73 on 17 df (p = 7e-05). DSG2 (HR = 1.238, p = 0.012), EIF6 (HR = 1.387, p = 0.046), and EXO1 (HR = 1.257, p = 0.047) all have hazard ratios greater than 1, indicating that higher expression of these genes is associated with poorer survival outcomes.

thumbnail
Download:
Fig 12. Multivariate Cox regression analysis identifying independent prognostic factors.

https://doi.org/10.1371/journal.pone.0349820.g012

Validation of prognostic value in an independent cohort

To validate the prognostic relevance of DSG2, EIF6, and EXO1 in our risk model, we utilized the independent dataset GSE13213 through the SurvExpress platform [33]. Cox regression model used to assess DSG2, EIF6, and EXO1 included only the expression levels of these three genes. The analysis revealed that all three genes maintained a significant association with patient prognosis in this external cohort. Specifically, DSG2 (coefficient: 0.2795, p-value: 0.05), EIF6 (coefficient: 0.8356, p-value: 0.019), and EXO1 (coefficient: 0.7033, p-value: 0.001) were all positively correlated with poor survival outcomes. The overall model performance was statistically robust, as indicated by a Log-Rank p-value < 0.001 and a hazard ratio (HR) of 3.19, underscoring the model’s strong prognostic power in stratifying patients.

Further stratification of the cohort into high-risk and low-risk groups based on the risk scores revealed a significant survival difference, with patients in the high-risk group exhibiting markedly shorter survival times compared to those in the low-risk group (Fig 13A). Moreover, expression analysis confirmed that DSG2, EIF6, and EXO1 were significantly upregulated in the high-risk group, consistent with our primary findings (Fig 13B). These validation results provide strong support for the clinical applicability of the proposed gene signature and reinforce the role of these three genes as potential biomarkers for risk stratification and prognosis in lung cancer.

thumbnail
Download:
Fig 13. Validation of prognostic gene signature using independent cohort (GSE13213).

A: Survival Analysis Based on Risk Groups in GSE13213 Dataset utilizing SurvExpress platform. B: Expression Levels of DSG2, EIF6, and EXO1 in High- and Low-Risk Groups.

https://doi.org/10.1371/journal.pone.0349820.g013

Discussion

Lung adenocarcinoma remains one of the leading causes of cancer-related mortality worldwide, with hypoxia recognized as a key contributor to tumor progression, metastasis, and therapy resistance. In this study, we comprehensively analyzed hypoxia-related differentially expressed genes (HRDEGs) in LUAD and identified 283 upregulated and 322 downregulated HRDEGs, reflecting the broad impact of hypoxic signaling on multiple cellular processes. Functional enrichment and protein-protein interaction analyses revealed that upregulated genes were primarily involved in cancer-related and cellular signaling pathways, whereas downregulated genes were enriched in immunity-associated pathways, highlighting the dual role of hypoxia in promoting tumor growth while modulating immune responses.

The enrichment of cell cycle–related genes, including PCNA, MCM7, CDKN2A, PRKDC, HDAC1, CDC6, CDK4, CHEK1, and E2F3, highlights a pronounced dysregulation of cell cycle control mechanisms in lung cancer. These genes are central to DNA replication, cell cycle progression, and checkpoint regulation, ensuring genomic integrity under normal physiological conditions [35–37]. Their coordinated upregulation suggests enhanced proliferative capacity and impaired checkpoint surveillance, which are hallmark features of tumorigenesis. In particular, aberrant activation of CDKs and E2F transcription factors can drive uncontrolled entry into the S phase [38], while altered expression of DNA damage response genes such as CHEK1 and PRKDC may allow cancer cells to bypass critical repair checkpoints, thereby promoting genomic instability [39,40]. Collectively, these findings underscore the pivotal role of cell cycle dysregulation in lung cancer development and progression.

In addition to cell cycle alterations, the HIF-1 signaling pathway (hsa04066) emerged as another significantly enriched pathway, with notable upregulation of genes including LDHA, EGLN3, SLC2A1, EIF4EBP1, PGK1, TIMP1, ENO1, ALDOA, GAPDH, PFKP, and PDK1. These genes are key mediators of cellular adaptation to hypoxic conditions and are closely associated with metabolic reprogramming in cancer. The elevated expression of glycolytic enzymes and glucose transporters, such as SLC2A1 and LDHA, reflects a shift toward aerobic glycolysis (the Warburg effect), enabling tumor cells to sustain rapid growth in oxygen-deprived microenvironments [41]. Taken together, the concurrent activation of cell cycle and HIF-1 signaling pathways suggests a coordinated mechanism by which lung cancer cells promote unchecked proliferation while adapting to hypoxic stress. These pathways may act synergistically to facilitate tumor progression and aggressiveness, highlighting their potential as therapeutic targets and prognostic biomarkers in lung cancer.

Beyond the activation of oncogenic and metabolic programs, lung cancer was characterized by a marked suppression of immune-associated pathways, reflecting the establishment of an immunosuppressive tumor microenvironment [42]. The coordinated downregulation of signaling cascades involved in cytokine communication, inflammatory responses, and immune cell activation indicates a weakened immune network within tumors [43]. Furthermore, attenuation of pathways governing innate immune sensing and effector mechanisms suggests reduced recognition and clearance of malignant cells by immune surveillance systems [44]. Such widespread impairment of immune signaling is consistent with tumor-driven immune evasion, in which cancer cells actively modulate immune pathways to avoid immune-mediated destruction [45]. This immune suppression may not only facilitate tumor progression but also contribute to resistance to immune-based therapies, underscoring the critical role of immune pathway dysregulation in lung cancer pathogenesis [43]. Interestingly, we have previously shown that HIF-1 suppresses NF-κB signaling [46], one of the pathways found to be significantly downregulated in the analysis of lung adenocarcinoma presented here.

Moreover, the top upregulated hub genes, including, MMP9 and SPP1 play key roles in extracellular matrix remodeling and metastasis [47,48], while CDH1 is important for maintaining cell-cell adhesion [49]. Other upregulated hub genes, including SOX2 and EZH2 regulate stem cell-like properties and epigenetic modifications, contributing to cancer stem cell maintenance and progression [50,51]. CDK4 and CDKN2A are crucial regulators of the cell cycle, promoting cell division and tumor growth [52,53]. BRCA1 and HDAC1 are involved in DNA repair and chromatin remodeling [54,55], while CXCL10 and TIMP1 are associated with immune responses [56,57]. The upregulation of these hub genes highlights their critical roles in tumor progression and their potential as therapeutic targets in lung cancer. On the other hand, the downregulation of IL6, TNF, and IL1B suggests impaired inflammatory signaling [58], while TLR4, FOS, and ICAM1 are involved in immune cell activation and adhesion [59,60], indicating potential immune evasion mechanisms. PECAM1 and KDR are associated with endothelial cell function and angiogenesis, both crucial for tumor vasculature [60,61]. The downregulation of PPARG, ITGAM, and CXCL12 further highlights disruptions in immune regulation and cell migration [62–64]. Overall, the downregulation of these hub genes suggests a weakened immune response, contributing to tumor progression and immune evasion in lung cancer.

Some of the survival associated genes are involved in the several cancers. For example, Li et al. reported that both the transcriptional and post-transcriptional expression levels of ADAM12 are significantly elevated in tumor tissues compared to normal tissues. Furthermore, increased ADAM12 expression was found to be associated with poor prognosis in patients with lung adenocarcinoma [65]. Ma et al. found that CCNE1 is upregulated in LUAD and that its knockdown suppresses tumor growth and metastasis. High CCNE1 expression correlates with advanced stage, larger tumors, lymph node involvement, and poor prognosis [66]. The downregulated group, categorized based on median expression levels, is associated with worse survival outcomes. Lower expression of these genes correlates with poorer prognosis, suggesting that their downregulation may contribute to tumor progression and aggressive cancer behavior [67,68]. These findings highlight the potential role of these downregulated genes in worsening survival outcomes and underscore their relevance as prognostic biomarkers for lung cancer.

Three-gene signature with prognostic value in LUAD

By integrating survival analysis and regression modeling, we further identified DSG2, EIF6, and EXO1 as independent prognostic factors, establishing a novel hypoxia-related gene signature that may inform both patient prognosis and therapeutic strategies in LUAD. These genes are associated with different aspects of biology but have previously been associated with hypoxia responses [69–71]. DSG2 is ubiquitously expressed and is a type of cadherin involved in intermediate filaments, controlling cell-to-cell junctions [72]. Its deregulation is also associated with numerous diseases such as cancer and autoimmune disease [73]. DSG2 expression is significantly associated with poor prognosis and reduced immune cell infiltration in various cancers [74]. EIF6 encodes a translation factor that controls association of 40S and 60S, is needed for the synthesis of 60S subunits and for translation [75]. It has been shown to be overexpressed in several human cancers [74]. EXO1 is an exonuclease involved in several important biological processes such as DNA damage repair, DNA replication and meiosis [76,77]. EXO1 has been proposed as a biomarker with both prognostic and diagnostic value across multiple cancers, highlighting its potential as a therapeutic target and its role in immunomodulation within the tumor microenvironment [77,78].

While numerous studies have reported associations between hypoxia-related genes and LUAD prognosis [18–20], our unbiased and integrative analysis identified this novel three-gene signature (DSG2, EIF6, and EXO1), which has not been previously reported as a prognostic panel. This signature captures distinct aspects of hypoxia-driven tumor biology: DSG2 regulates cell adhesion and tumor progression [69], EIF6 controls translation initiation supporting rapid tumor growth under hypoxia [70], and EXO1 mediates DNA repair under replication stress, which is often exacerbated in hypoxic tumors [71]. Multiple studies have demonstrated that EXO1 is significantly overexpressed in LUAD tissues compared with adjacent normal lung tissues and that high EXO1 expression is associated with poorer overall survival and more aggressive clinicopathological features, including advanced stage and increased tumor migration [79]. Genetic studies have demonstrated that variants in key hypoxia signaling pathways, including EXO1, are associated with poor overall survival in non–small cell lung cancer, indicating a connection between hypoxia‑driven stress responses and EXO1 expression and function [80]. EIF6 has been implicated in adaptive translational control under hypoxic stress, reflecting a mechanism by which tumor cells maintain protein synthesis in low-oxygen environments, supporting survival and growth in hypoxic regions of LUAD tumors [81]. Similarly, bioinformatic analyses have identified EIF6 overexpression in LUAD, where elevated EIF6 correlates with enhanced tumor cell proliferation, invasion, and worse patient prognosis, suggesting its role in translational regulation and tumor progression in lung cancer [82]. Although direct studies linking DSG2 to hypoxia are limited, DSG2 has been reported to be upregulated in LUAD and associated with tumor progression and metastatic potential, which is consistent with its identification as an adverse prognostic factor in our study [83]. Importantly, these genes were validated in an independent cohort and may have clinical implications, as hypoxia influences the tumor microenvironment and response to immune checkpoint inhibitors. Thus, the DSG2-EIF6-EXO1 signature not only provides robust prognostic stratification but also offers mechanistic and potentially therapeutic insight, distinguishing it from previously reported hypoxia-related prognostic markers. Collectively, this evidence supports the view that DSG2, EIF6, and EXO1 are not only relevant to tumor aggressiveness in LUAD but also fit within a hypoxia‑centered framework, consistent with their upregulation in more aggressive, hypoxic tumor regions and their association with poor survival in LUAD patients.

Comparison with previous biomarker LUAD studies

Several studies have explored hypoxia-related prognostic biomarkers in LUAD, but their scope and methodology differ from our approach. Li et al. developed a 7-gene immune–hypoxia signature. Their focus was on genes at the intersection of immune and hypoxia pathways [18]. In contrast, our study analyzed a genome-wide set of hypoxia-related mRNAs, identifying three independent prognostic hub genes (DSG2, EIF6, EXO1) that do not overlap with the Li et al. signature. Additionally, we validated our findings in independent LUAD cohorts, strengthening their clinical relevance and providing a distinct prognostic framework centered on core hypoxia-driven mRNAs. Xiong et al. focused on 15 hypoxia-related genes to classify hypoxia clusters and constructed a ceRNA-based network [19], emphasizing lncRNAs and clustering analyses. Our approach differs in analyzing a broader hypoxia-related gene set, integrating protein–protein interaction networks, and identifying independent prognostic hub genes. This allows for a more systematic understanding of hypoxia-driven mechanisms and provides stronger translational relevance for LUAD prognosis. Another research group established a 9-lncRNA hypoxia signature [20]. While informative, their study focused exclusively on non-coding RNAs. In contrast, our study emphasizes protein-coding mRNAs, identifying genes with independent prognostic value and mechanistic relevance. Our findings complement the lncRNA-based signature by highlighting additional molecular targets that can improve prognosis prediction and deepen mechanistic insights into hypoxia-driven LUAD progression. Overall, our study extends prior work by providing a comprehensive, multilayered analysis of hypoxia-related genes, integrating functional enrichment, PPI networks, and survival modeling to identify independent prognostic hub genes. This approach not only captures broader hypoxia-induced changes but also establishes a distinct prognostic model for LUAD.

Limitations of this study

Despite the independent cohort validation and the novelty of our three-gene signature (DSG2, EIF6, and EXO1), several limitations should be acknowledged. First, the study relies on public retrospective datasets, which may introduce selection bias and limit generalizability. Second, there is cross-platform heterogeneity between datasets, including potential batch effects and annotation differences that could affect results. Third, the curated hypoxia-related gene sets used may not capture all relevant hypoxia-associated genes, potentially omitting novel or unannotated candidates. Fourth, the study lacks protein-level validation to confirm whether mRNA expression correlates with protein abundance and function. Fifth, no experimental validation using LUAD cell lines or patient-derived samples was performed, limiting mechanistic interpretation. Finally, the validation relied on an online platform (SurvExpress) rather than a fully reproduced external analytic pipeline, which may affect reproducibility.

To address these limitations, future work should include experimental validation of DSG2, EIF6, and EXO1 in LUAD models and patient samples, including qPCR or protein-level assays. Additional studies should aim to develop therapeutic strategies targeting high-risk patients identified by this signature. Moreover, efforts to harmonize datasets across platforms and expand hypoxia-related gene curation could improve the robustness and translational relevance of the findings.

Conclusions

In conclusion, our analysis identifies DSG2, EIF6, and EXO1 as key hypoxia-related genes with significant prognostic value in lung adenocarcinoma. These genes, through their involvement in hypoxia-related pathways, may play critical roles in tumor progression and patient survival. The findings suggest that DSG2, EIF6, and EXO1 could serve as potential biomarkers for predicting patient outcomes and may offer novel therapeutic targets for improving prognosis in lung adenocarcinoma. Further validation of these genes in larger cohorts and clinical settings is necessary to confirm their utility in clinical practice.

Supporting information

S1 File. Supplementary Tables S1-S9.

S1. List of collected hypoxia related genes; Table S2. List of DEGs in TCGA LUAD; Table S3. List of DEGs in GSE18842; Table S4. List of hypoxia related DEGs in both datasets; Table S5. Enriched pathways associated with the upregulated genes (FDR < 0.05); Table S6. Enriched pathways associated with the downregulated genes (FDR < 0.05); Table S7. Rank of the PPI network by Degree method; Table S8. Significant Genes Identified in Kaplan-Meier Survival Analysis; Table S9. Significant Genes Identified in Univariate regression Analysis.

https://doi.org/10.1371/journal.pone.0349820.s001

(XLSX)

Acknowledgments

We thank all the members of the scientific community for making their datasets public.

References

  1. 1. Zhang Y, Vaccarella S, Morgan E, Li M, Etxeberria J, Chokunonga E, et al. Global variations in lung cancer incidence by histological subtype in 2020: a population-based study. Lancet Oncol. 2023;24(11):1206–18. pmid:37837979
  2. 2. Seguin L, Durandy M, Feral CC. Lung adenocarcinoma tumor origin: a guide for personalized medicine. Cancers (Basel). 2022;14(7):1759. pmid:35406531
  3. 3. Zhang C, Liu Y, Lu Y, Chen Z, Liu Y, Mao Q, et al. Identification of potential biomarkers for lung adenocarcinoma: a study based on bioinformatics analysis combined with validation experiments. Front Oncol. 2024;14:1425895. pmid:39364312
  4. 4. Emami Nejad A, Najafgholian S, Rostami A, Sistani A, Shojaeifar S, Esparvarinha M, et al. The role of hypoxia in the tumor microenvironment and development of cancer stem cell: a novel approach to developing treatment. Cancer Cell Int. 2021;21(1):62. pmid:33472628
  5. 5. Ciepła J, Smolarczyk R. Tumor hypoxia unveiled: insights into microenvironment, detection tools and emerging therapies. Clin Exp Med. 2024;24(1):235. pmid:39361163
  6. 6. Petrova V, Annicchiarico-Petruzzelli M, Melino G, Amelio I. The hypoxic tumour microenvironment. Oncogenesis. 2018;7(1):10. pmid:29362402
  7. 7. Kenneth NS, Rocha S. Regulation of gene expression by hypoxia. Biochem J. 2008;414(1):19–29. pmid:18651837
  8. 8. Regulation of gene transcription by hypoxia-inducible factor 1α. 2021. pp. 480–9.
  9. 9. Bae T, Hallis SP, Kwak M-K. Hypoxia, oxidative stress, and the interplay of HIFs and NRF2 signaling in cancer. Exp Mol Med. 2024;56(3):501–14. pmid:38424190
  10. 10. Bui BP, Nguyen PL, Lee K, Cho J. Hypoxia-inducible factor-1: a novel therapeutic target for the management of cancer, drug resistance, and cancer-related pain. Cancers. 2022;14:6054.
  11. 11. Gonzalez-Avila G, Sommer B, Flores-Soto E, Aquino-Galvez A. Hypoxic effects on matrix metalloproteinases’ expression in the tumor microenvironment and therapeutic perspectives. Int J Mol Sci. 2023;24:16887.
  12. 12. Frost J, Frost M, Batie M, Jiang H, Rocha S. Roles of HIF and 2-oxoglutarate-dependent dioxygenases in controlling gene expression in hypoxia. Cancers (Basel). 2021;13(2):350. pmid:33477877
  13. 13. Druker J, Wilson JW, Child F, Shakir D, Fasanya T, Rocha S. Role of hypoxia in the control of the cell cycle. Int J Mol Sci. 2021;22(9):4874. pmid:34062959
  14. 14. Collier H, Albanese A, Kwok C-S, Kou J, Rocha S. Functional crosstalk between chromatin and hypoxia signalling. Cell Signal. 2023;106:110660. pmid:36990334
  15. 15. Roy S, Kumaravel S, Sharma A, Duran CL, Bayless KJ, Chakraborty S. Hypoxic tumor microenvironment: implications for cancer therapy. Exp Biol Med (Maywood). 2020;245(13):1073–86. pmid:32594767
  16. 16. Chen Z, Han F, Du Y, Shi H, Zhou W. Hypoxic microenvironment in cancer: molecular mechanisms and therapeutic interventions. Signal Transduct Target Ther. 2023;8(1):70. pmid:36797231
  17. 17. Matic N, Pettersson L, Sellebjerg F, Lindberg L, Roberg K, Wiechec E. Prognostic value of hypoxia-responsive gene expression profile in patients diagnosed with head and neck squamous cell carcinoma. Transl Oncol. 2024;39:101841. pmid:38016355
  18. 18. Li Y, Huang H, Jiang M, Yu N, Ye X, Huang Z, et al. Identification and validation of a hypoxia-immune signature for overall survival prediction in lung adenocarcinoma. Front Genet. 2022;13:975279. pmid:36263421
  19. 19. Xiong L, He X, Wang L, Dai P, Zhao J, Zhou X, et al. Hypoxia-associated prognostic markers and competing endogenous RNA coexpression networks in lung adenocarcinoma. Sci Rep. 2022;12(1):21340. pmid:36494419
  20. 20. Zhang J, Kondowe B, Zhang H, Xie X, Song Q, Guo B, et al. Identification of prognostic indicator based on hypoxia-related lncRNAs analysis in lung adenocarcinoma. Malawi Med J. 2024;36(3):170–8. pmid:40018402
  21. 21. Rahal Z, El Darzi R, Moghaddam SJ, Cascone T, Kadara H. Tumour and microenvironment crosstalk in NSCLC progression and response to therapy. Nat Rev Clin Oncol. 2025;22(7):463–82. pmid:40379986
  22. 22. Shi R, Sun J, Zhou H, Hu T, Gao Z, Wang X, et al. Hypoxia within tumor microenvironment characterizes distinct genomic patterns and aids molecular subtyping for guiding individualized immunotherapy. J Big Data. 2024;11(1).
  23. 23. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2012;2(5):401–4. pmid:22588877
  24. 24. Sanchez-Palencia A, Gomez-Morales M, Gomez-Capilla JA, Pedraza V, Boyero L, Rosell R, et al. Gene expression profiling reveals novel biomarkers in nonsmall cell lung cancer. Int J Cancer. 2011;129(2):355–64. pmid:20878980
  25. 25. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47. pmid:25605792
  26. 26. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A. 2005;102(43):15545–50. pmid:16199517
  27. 27. Huang DW, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009;37(1):1–13. pmid:19033363
  28. 28. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B: Stat Methodol. 1995;57(1):289–300.
  29. 29. Szklarczyk D, Gable AL, Lyon D, Junge A, Wyder S, Huerta-Cepas J, et al. STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res. 2019;47(D1):D607–13. pmid:30476243
  30. 30. Chin C-H, Chen S-H, Wu H-H, Ho C-W, Ko M-T, Lin C-Y. cytoHubba: identifying hub objects and sub-networks from complex interactome. BMC Syst Biol. 2014;8 Suppl 4(Suppl 4):S11. pmid:25521941
  31. 31. Therneau T. A package for survival analysis in R.
  32. 32. Friedman J, Hastie T, Tibshirani R. Regularization paths for generalized linear models via coordinate descent. J Stat Softw. 2010;33(1):1–22. pmid:20808728
  33. 33. Aguirre-Gamboa R, Gomez-Rueda H, Martínez-Ledesma E, Martínez-Torteya A, Chacolla-Huaringa R, Rodriguez-Barrientos A, et al. SurvExpress: an online biomarker validation tool and database for cancer gene expression data using survival analysis. PLoS One. 2013;8(9):e74250. pmid:24066126
  34. 34. Zhao Y, Xing C, Deng Y, Ye C, Peng H. HIF-1α signaling: essential roles in tumorigenesis and implications in targeted therapies. Genes Dis. 2023;11(1):234–51. pmid:37588219
  35. 35. Barnum KJ, O’Connell MJ. Cell cycle regulation by checkpoints. Methods Mol Biol. 2014;1170:29–40. pmid:24906307
  36. 36. Pedroza-Garcia JA, Xiang Y, De Veylder L. Cell cycle checkpoint control in response to DNA damage by environmental stresses. Plant J. 2022;109(3):490–507. pmid:34741364
  37. 37. Rankin BD, Rankin S. The MCM2-7 complex: roles beyond DNA unwinding. Biology (Basel). 2024;13(4):258. pmid:38666870
  38. 38. Bertoli C, Herlihy AE, Pennycook BR, Kriston-Vizi J, de Bruin RAM. Sustained E2F-dependent transcription is a key mechanism to prevent replication-stress-induced DNA damage. Cell Rep. 2016;15(7):1412–22. pmid:27160911
  39. 39. Zhang Y, Wen G-M, Wu C-A, Jing Z-L, Li D-Z, Liu G-L, et al. PRKDC is a prognostic marker for poor survival in gastric cancer patients and regulates DNA damage response. Pathol Res Pract. 2019;215(8):152509. pmid:31255330
  40. 40. Klomp JE, Lee YS, Goodwin CM, Papke B, Klomp JA, Waters AM, et al. CHK1 protects oncogenic KRAS-expressing cells from DNA damage and is a target for pancreatic cancer treatment. Cell Rep. 2021;37(9):110060. pmid:34852220
  41. 41. Ooi AT, Gomperts BN. Molecular pathways: targeting cellular energy metabolism in cancer via inhibition of SLC2A1 and LDHA. Clin Cancer Res. 2015;21(11):2440–4. pmid:25838393
  42. 42. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell. 2011;144(5):646–74. pmid:21376230
  43. 43. Chen DS, Mellman I. Oncology meets immunology: the cancer-immunity cycle. Immunity. 2013;39(1):1–10. pmid:23890059
  44. 44. Schreiber RD, Old LJ, Smyth MJ. Cancer immunoediting: integrating immunity’s roles in cancer suppression and promotion. Science. 2011;331(6024):1565–70. pmid:21436444
  45. 45. Herbst RS, Morgensztern D, Boshoff C. The biology and management of non-small cell lung cancer. Nature. 2018;553(7689):446–54. pmid:29364287
  46. 46. Bandarra D, Biddlestone J, Mudie S, Müller H-AJ, Rocha S. HIF-1α restricts NF-κB-dependent gene expression to control innate immunity signals. Dis Model Mech. 2015;8(2):169–81. pmid:25510503
  47. 47. Pang X, Xie R, Zhang Z, Liu Q, Wu S, Cui Y. Identification of SPP1 as an extracellular matrix signature for metastatic castration-resistant prostate cancer. Front Oncol. 2019;9:924. pmid:31620371
  48. 48. Napoli S, Scuderi C, Gattuso G, Bella VD, Candido S, Basile MS, et al. Functional roles of matrix metalloproteinases and their inhibitors in melanoma. Cells. 2020;9:1151.
  49. 49. Bhattacharyya S, Mote RD, Freimer JW, Tiwari M, Bansi Singh S, Arumugam S, et al. Cell-cell adhesions in embryonic stem cells regulate the stability and transcriptional activity of β-catenin. FEBS Lett. 2022;596(13):1647–60. pmid:35344589
  50. 50. French R, Pauklin S. Epigenetic regulation of cancer stem cell formation and maintenance. Int J Cancer. 2021;148(12):2884–97. pmid:33197277
  51. 51. Mirzaei S, Paskeh MDA, Entezari M, Mirmazloomi SR, Hassanpoor A, Aboutalebi M, et al. SOX2 function in cancers: association with growth, invasion, stemness and therapy response. Biomed Pharmacother. 2022;156:113860. pmid:36272267
  52. 52. Baker SJ, Poulikakos PI, Irie HY, Parekh S, Reddy EP. CDK4: a master regulator of the cell cycle and its role in cancer. Genes Cancer. 2022;13:21–45. pmid:36051751
  53. 53. Choi SA, Moon YJ, Koh EJ, Phi JH, Lee JY, Kim KH, et al. Cyclin-dependent kinase inhibitor 2A is a key regulator of cell cycle arrest and senescence in endothelial colony-forming cells in Moyamoya disease. J Korean Neurosurg Soc. 2023;66(6):642–51. pmid:37138505
  54. 54. Yarden RI, Brody LC. BRCA1 interacts with components of the histone deacetylase complex. Proc Natl Acad Sci U S A. 1999;96(9):4983–8. pmid:10220405
  55. 55. Bhaskara S. Histone deacetylases 1 and 2 regulate DNA replication and DNA repair: potential targets for genome stability-mechanism-based therapeutics for a subset of cancers. Cell Cycle. 2015;14(12):1779–85. pmid:25942572
  56. 56. Allen JR, Ge L, Huang Y, Brauer R, Parimon T, Cassel SL, et al. TIMP-1 promotes the immune response in influenza-induced acute lung injury. Lung. 2018;196(6):737–43. pmid:30167842
  57. 57. Elemam NM, Talaat IM, Maghazachi AA. CXCL10 chemokine: a critical player in RNA and DNA viral infections. Viruses. 2022;14(11):2445. pmid:36366543
  58. 58. Schultheiß C, Willscher E, Paschold L, Gottschick C, Klee B, Henkes S-S, et al. The IL-1β, IL-6, and TNF cytokine triad is associated with post-acute sequelae of COVID-19. Cell Rep Med. 2022;3(6):100663. pmid:35732153
  59. 59. Zheng M, Ambesi A, McKeown-Longo PJ. Role of TLR4 receptor complex in the regulation of the innate immune response by fibronectin. Cells. 2020;9(1):216. pmid:31952223
  60. 60. Guerra-Espinosa C, Jiménez-Fernández M, Sánchez-Madrid F, Serrador JM. ICAMs in immunity, intercellular adhesion and communication. Cells. 2024;13:339.
  61. 61. del Puerto-Nevado L, Rojo F, Zazo S, Caramés C, Rubio G, Vega R, et al. Active angiogenesis in metastatic renal cell carcinoma predicts clinical benefit to sunitinib-based therapy. Br J Cancer. 2014;110(11):2700–7. pmid:24786599
  62. 62. Angeli V, Hammad H, Staels B, Capron M, Lambrecht BN, Trottein F. Peroxisome proliferator-activated receptor gamma inhibits the migration of dendritic cells: consequences for the immune response. J Immunol. 2003;170(10):5295–301. pmid:12734379
  63. 63. Guyon A. CXCL12 chemokine and its receptors as major players in the interactions between immune and nervous systems. Front Cell Neurosci. 2014;8:65. pmid:24639628
  64. 64. Huang H, Wang Q, Ma L, Wu Y. ITGAM: a pivotal regulator in macrophage dynamics and cardiac function during sepsis-induced cardiomyopathy. Cureus. 2024;16:e59342.
  65. 65. Li K, Quan L, Huang F, Li Y, Shen Z. ADAM12 promotes the resistance of lung adenocarcinoma cells to EGFR-TKI and regulates the immune microenvironment by activating PI3K/Akt/mTOR and RAS signaling pathways. Int Immunopharmacol. 2023;122:110580. pmid:37418984
  66. 66. Ma G, Yang L, Dong J, Zhang L. CCNE1 promotes progression and is associated with poor prognosis in lung adenocarcinoma. Curr Pharm Biotechnol. 2022;23(9):1168–78. pmid:34792009
  67. 67. Cizkova M, Vacher S, Meseure D, Trassard M, Susini A, Mlcuchova D, et al. PIK3R1 underexpression is an independent prognostic marker in breast cancer. BMC Cancer. 2013;13:545. pmid:24229379
  68. 68. Wei X, Chen L, Yang A, Lv Z, Xiong M, Shan C. ADRB2 is a potential protective gene in breast cancer by regulating tumor immune microenvironment. Transl Cancer Res. 2021;10(12):5280–94. pmid:35116377
  69. 69. Myo Min KK, Ffrench CB, McClure BJ, Ortiz M, Dorward EL, Samuel MS, et al. Desmoglein-2 as a cancer modulator: friend or foe? Front Oncol. 2023;13:1327478. pmid:38188287
  70. 70. Sun L, Liu S, Wang X, Zheng X, Chen Y, Shen H. eIF6 promotes the malignant progression of human hepatocellular carcinoma via the mTOR signaling pathway. J Transl Med. 2021;19(1):216. pmid:34016142
  71. 71. García-Rodríguez N, Domínguez-García I, Domínguez-Pérez MDC, Huertas P. EXO1 and DNA2-mediated ssDNA gap expansion is essential for ATR activation and to maintain viability in BRCA1-deficient cells. Nucleic Acids Res. 2024;52(11):6376–91. pmid:38721777
  72. 72. Huber O. Structure and function of desmosomal proteins and their role in development and disease. Cell Mol Life Sci. 2003;60(9):1872–90. pmid:14523549
  73. 73. Hegazy M, Perl AL, Svoboda SA, Green KJ. Desmosomal cadherins in health and disease. Annu Rev Pathol. 2022;17:47–72. pmid:34425055
  74. 74. Zhang G, Chen Z, Wang Y, Huang A, Nie F, Gao L, et al. Up-regulated DSG2 promotes tumor growth and reduces immune infiltration in cervical cancer. Pathol Res Pract. 2024;262:155554. pmid:39226803
  75. 75. Hou S, Tang H, Dong Z, Zhou W. eIF6: a promising therapeutic target for gastric carcinoma via PI3K/AKT pathway modulation. World J Surg Oncol. 2025;23(1):113. pmid:40170052
  76. 76. Schaetzlein S, Chahwan R, Avdievich E, Roa S, Wei K, Eoff RL, et al. Mammalian Exo1 encodes both structural and catalytic functions that play distinct roles in essential biological processes. Proc Natl Acad Sci U S A. 2013;110(27):E2470–9. pmid:23754438
  77. 77. Yan S, Gao S, Zhou P. Multi-functions of exonuclease 1 in DNA damage response and cancer susceptibility. Radiat Med Prot. 2021;2:146–54.
  78. 78. Liu Z, Huang S, Luo R, Shi X, Xiu M, Wang Y, et al. EXO1’s pan-cancer roles: diagnostic, prognostic, and immunological analyses through bioinformatics. Discov Oncol. 2025;16(1):310. pmid:40074873
  79. 79. Zhou C-S, Feng M-T, Chen X, Gao Y, Chen L, Li L-D, et al. Exonuclease 1 (EXO1) is a potential prognostic biomarker and correlates with immune infiltrates in lung adenocarcinoma. Onco Targets Ther. 2021;14:1033–48. pmid:33623391
  80. 80. Li R, Gu J, Heymach JV, Shu X, Zhao L, Han B, et al. Hypoxia pathway genetic variants predict survival of non-small-cell lung cancer patients receiving platinum-based chemotherapy. Carcinogenesis. 2017;38(4):419–24. pmid:28186269
  81. 81. Yi T, Papadopoulos E, Hagner PR, Wagner G. Hypoxia-inducible factor-1α (HIF-1α) promotes cap-dependent translation of selective mRNAs through up-regulating initiation factor eIF4E1 in breast cancer cells under hypoxia conditions. J Biol Chem. 2013;288(26):18732–42. pmid:23667251
  82. 82. Shen Y, Zhang R, Li X. Identification of eIF6 as a prognostic factor that drives tumor progression and predicts arsenic trioxide efficacy in lung adenocarcinoma. Mol Biol Rep. 2023;50(2):1167–80. pmid:36435920
  83. 83. Wang C-C, Lin J-Y, Wang C-Y, Shen W-J, Liao P-C, Ho Y-F, et al. DSG2 attenuates gemcitabine efficacy through PTX3 in lung adenocarcinoma. Biochim Biophys Acta Mol Basis Dis. 2025;1871(6):167881. pmid:40316058