A P4HA2-dominated hypoxia signature enables response stratification of multi-therapeutic response in cervical cancer

A P4HA2-dominated hypoxia signature enables response stratification of multi-therapeutic response in cervical cancer | 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 A P4HA2-dominated hypoxia signature enables response stratification of multi-therapeutic response in cervical cancer Xiaojiao Li, Xinyuan Zhang, Yilin Dai, Fanwei Huang, Rui Wei, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6832842/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 21 Jan, 2026 Read the published version in BMC Cancer → Version 1 posted 11 You are reading this latest preprint version Abstract Background Cervical cancer, especially cervical squamous cell carcinoma (CSCC), is a major public health issue in low - and middle - income countries, with advanced recurrence and metastasis linked to poor prognosis. It shows significant intra - tumoral phenotypic heterogeneity and plasticity. Methods We analyzed the single-cell RNA sequencing data of cervical squamous cell carcinoma available in the Comprehensive Gene Expression Database (GEO) and identified the tumor cell subtype exhibiting hypoxic characteristics. We extracted differentially expressed genes (HRDEGs) between this hypoxia-related cluster and other tumor cells. Based on the CSCC bulk RNA sequencing data published in the Cancer Genome Atlas (TCGA), this subtype was identified to be closely associated with poor prognosis in CSCC.101 combinations consisting of 10 machine learning were used for screening prognostic biomarkers in HRDEGs, and a hypoxia signature was established by multivariate COX regression. Results The hypoxia signature was validated using the GEO external database. Correlation analysis identified the hypoxia signature as significantly associated with hypoxia and tumor invasion, and verified that higher hypoxia signature are closely related to poorer immune infiltration and responses of immunotherapy and chemotherapy. In addition, the key gene P4HA2 in the hypoxia signature has been demonstrated to be associated with the malignant phenotypes of tumor cells and the regulation of HIF-1α stability. Conclusions Overall, this hypoxia signature is a promising independent prognostic factor, provides new biomarkers for the prognosis of CSCC and a good reference for personalized and precision medicine. cervical squamous cell carcinoma hypoxia prognosis scRNA-seq Bulk RNA-seq HIF-1α machine learning Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Cervical cancer (CC) was the fourth most common cancer in women world widely[1, 2]. Although the system implementation of HPV vaccines has reduced the incidence and mortality of cancer substantially, the prognosis for patients with advanced-stage, recurrent, and metastatic CC remains poor[3, 4]. Estimated proportions of 37% CC were diagnosed with locally advanced disease which were standardly treated with concurrent chemoradiotherapy (CCRT)[5], among which, around 30% of post-treatment patients would progress into recurrence or subsequent metastases[6]. Additionally, more than 10% CC cases were made up with stage IVB disease and poses an even more challenging treatment scenario[7–11].Neoadjuvant chemotherapy (NACT) followed by radical hysterectomy has been considered an alternative to CCRT for stage IVB patients but with a major drawback of delaying chemo-resistant patient from on-time local treatment[6]. Early risk stratification and intervention play crucial roles in allowing patients to be treated precisely and promptly, and accordingly leading to better outcomes. A promising approach is to identify and validate novel biomarkers with prognostic and predictive significance offering insights into targetable molecular, thereby facilitating personalized cancer treatment[12]. CC heterogeneity is one of the major challenges of translating biomarker driven therapy into clinical practice[12]. With the advancement of recent genomic, transcriptomic and proteomic technologies, cervical squamous cell carcinoma (CSCC) is characterized with high tumor heterogeneity, accounts for approximately 80% of all CC[1, 13]. The diversity of squamous epithelial cells in CSCC encompasses of morphology, transcriptomic profiles, epigenomic modifications, metabolic activity, and accordingly treatment responses[14, 15]. Within the heterogenous solid tumor microenvironment, the propensity of CSCC for hypoxia is widely revealed, which is highly involved in responsiveness to radiotherapy, chemotherapy, and immunotherapy [16–18]. In contrast to normal uterine cervix tissue consuming oxygen, CSCC is depicted with hypoxia-glycolysis-acidosis paradigm due to the perturbation and deficiency in microcirculation[19]. Conversely, hypoxia itself also renders tumor heterogeneity simply by its condition of inequivalent oxygen distribution[20]. This oxygen deprivation acts indeed as a surviving stressor compelling cancer cells to adapt to this unfavorable energy condition by prompting variety of genes as p53, HIF1a or GLUT1 to evade apoptosis or necrosis [21]. Along with this adaptation, cancer cells re-allocate its energy turning out to mediate various essential processes such as proliferation, and angiogenesis, promoting tumor invasion, metastasis and treatment resistance in terms of chemotherapy, radiotherapy, and immunotherapy [22–24]. HIF1a is now elucidated as an important regulator of tumor survival via manipulating hypoxic cascades. Therefore, efforts are extensively underway to decipher the role of hypoxia in CSCC to understand the tumor evolution and immune activity modulation, and consequently, improve the patient outcomes. Recent studies, with the applications of high-throughput-sequencing technologies, the role of hypoxia on cancer cell heterogeneity has been elucidated in a new era with single cell resolution. Guo C. et al. presented hypoxia role in modulating macrophage polarization in 2 CCSC samples[25]. Qiu J. et al. unveiled a hypoxia related clusters as a progenitor of CC by preforming a comprehensive single-cell atlas of totally 17 samples including 3 CCSC cases[26]. Although these studies have proposed elaborate molecular mechanisms supporting tumor survival and carcinogenesis, the case sizes enrolled in these studies were usually limited with absence of follow-ups. Integration of single-cell sequencing and bulk RNA sequencing could constitute a good investigation strategy to broaden single-cell findings to larger populations. By applying 10 machine learning algorithms into 101 combinations, we visualized a significant relationship between a P4HA2 dominated hypoxia signature and poor outcomes of 297 CSCC patients in silico . In this respect, we further conducted Transwell migration/invasion assays and CCK-8 assays to evaluate the role of P4HA2 with cancer cell proliferation, infiltration and aggressiveness in three cervical cell lines. Collectively, herein, we identified a hypoxia signature that presents clinical significance and provided mechanical evidences of the potential to target hypoxia and P4HA2 in treating CCSC. Materials and methods Data acquisition The scRNA-seq data GSE208653 and GSE197461 are publicly available datasets from a comprehensive gene expression database(GEO, https://www.ncbi.nlm.nih.gov/geo/ ). We removed samples of cervical adenocarcinoma, CIN, and normal cervical epithelium, and only retained 5 CSCC patient samples. We obtained RNA-seq data and clinical information of TCGA-CSCC (n = 246) and GSE52903 (n = 51), from the TCGA database and GEO dataset, respectively. In addition, datasets GSE100080, GSE7410, GSE26511, and GSE146114 were obtained from GEO to explore the correlation between risk scores and clinical staging. scRNAseq datasets and processing The "scanpy" package( https://scanpy.readthedocs.io/ ) in Python is used for quality control of scRNA seq data. We excluded cells with gene expression of less than 300 genes or more than 1000 genes, as well as cells with mitochondrial gene expression greater than 25%. We further normalized and scaled the original count, and then performed principal component analysis. [27]. Meanwhile, the "bbknn" package( https://github.com/Teichlab/bbknn ) was used to remove the batch effect of scRNA-seq raw data[28]. Using unsupervised clustering analysis, we identified 24 cell clusters and displayed them using UMAP. Then, we annotated each cell cluster based on known marker genes. Malignant cell identification The “inferCNVpy”package ( https://github.com/icbi-lab/infercnvpy ) was utilized in Python to infer the copy number variation (CNV) of epithelial cells in integrated single-cell RNA sequencing (scRNA-seq) data. In this analysis, T cells were used as a reference cell type to distinguish malignant epithelial cells from normal epithelial cells. Cell clusters with high CNV scores are considered malignant cells. Expression programs of intratumoral heterogeneity To depict tumor cells heterogeneity, we performed Non-negative matrix factorization (NMF) with the "geneNMF" R package( https://github.com/carmonalab/GeneNMF ) and determined the robust program with the k value was set between 3:9, and determined the consensus programs in metaprograms (MPs) that can be stably identified. In addition, We select the top 30 genes of each MP as the feature genes, and use the scanpy.tl.score_genes function to calculate the MP scores of each cell, and assign each cancer cell to the MP cluster with the highest score and project it onto the UMAP map. DEGs and Biological function and pathway enrichment analysis DEGs between MP7 and the other programs were obtained by the Wilcoxon rank-sum test, and excluded genes with logfoldchanges 0.01. The DEGs were further used for gene set enrichment analysis through KEGG (Kyoto Encyclopedia of Genes and Genomes) and GO (Gene Ontology) with “clusterProfiler” packages( https://github.com/YuLab-SMU/clusterProfiler ) in R[29]. Pathways with adjusted p-values < 0.05 are considered significantly enriched in MP7 Batch effect removal of Bulk-RNA datasets We Adopted the "SVA" package( https://bioconductor.org/packages/release/bioc/html/sva.html ) to eliminate batch effects of bulk RNA data[30]. Principal Component Analysis (PCA) was performed using "FactoMineR"( https://cran.r-project.org/web/packages/FactoMineR/index.html ) and visualized using the "Factoextra"( https://cran.r-project.org/web//packages/factoextra/index.html ) software package. Construction and validation of the risk model The coxph function of the “survival” package( https://github.com/therneau/survival ) was used to perform univariate Cox regression analysis, the prognosis-related genes were identified based on the criteria of p value < 0.01. We adopted a method that encompasses 101 combinations of machine learning which has been proven to robustly identify key genes for developing reliable hypoxia signatures, including Super Partial Correlation (SuperPC), Random Forest, Support Vector Machine (SVM), Least Absolute Shrinkage and Selection Operator (Lasso), Gradient Boosting Machine (GBM), Elastic Net, Stepwise Cox, Ridge, CoxBoost, and Partial Least Squares with Cox regression (plsRcox). Among them, RSF, LASSO, CoxBoost, StepCox bidirectional, and StepCox reverse were used to perform the first step of dimensionality reduction and variable screening[31]. The TCGA data was split into a training set and an internal validation set, while GSE52903 is considered as the external validation set, and the signature with the minimum number of genes and the best C-index was ultimately selected. HRPS = 0.9715×EGLN1 + 0.1442×ITGA5 − 0.7106×DAPK2 + 0.2035×PLOD2 + 0.2073×P4HA2 + 0.6229×AGPAT4. Evaluating the predictive value of the model We stratified patients according to the hypoxia signature and determined the optimal cutoff value using the surv_cutpoint function in the "survminer" package( https://github.com/kassambara/survminer ). Kaplan-Meier (KM) survival analysis was employed to assess the prognostic impact of the hypoxic signature. Additionally, ROC curves were utilized to validate the accuracy and stability of predicting 1-year, 3-year, and 5-year survival based on the hypoxia signature, which were generated through the "timeROC" package( https://github.com/cran/timeROC/blob/master/R/timeROC_3.R ). Clinical relevance of the hypoxia signature We conducted univariate and multivariate Cox regression analysis using the "survival" package to determine whether the hypoxia signature is an independent prognostic factor for CSCC patients, and visualized the results using the "forestplot" package( https://cran.r-project.org/package=forestplot ). Based on this, a nomogram of the TCGA cohort was constructed for clinical application including age, T, N, M, pathologic stage, and the hypoxia signature. Furthermore, to validate the association between the hypoxia signature and clinical staging with a larger sample size, in addition to the TCGA and GSE52903 cohorts, we included GSE100080, GSE7410, GSE146114, and GSE26511 cohorts. The relationship between Figo staging and hypoxia signature was visually demonstrated through boxplot generated by the “ggplot2” package( https://cran.r-project.org/package=ggplot2 ). Immune landscape analyses The "ESTIMATE" R package( https://estimate.r-forge.r-project.org/ ) was employed to analyze the stromal, immune, and ESTIMATE score of CSCC patients to quantify immune activation levels based on gene expression profiles. Furthermore, the assessment of the abundance of immune microenvironment and functions was conducted by the "IBOR" package( https://github.com/IOBR/IOBR ), which includes several bioinformatics algorithms like "MCP-counter," "IPS", "CIBERSORT," and "quanTIseq" algorithms[32]. Tumor mutation burden and drug response analyses SNP information was collected from the TCGA database, and then the mutation profiles was analyzed based on the risk stratification using the “maftools” R package( https://www.bioconductor.org/packages/release/bioc/html/maftools.html )[33]. We applied the tumor immune dysfunction and exclusion (TIDE) and the Immune phenotype score (IPS) to predict the potential immunotherapy responses in CSCC. IPS results for 20 solid tumors in TCGA are available on TCIA ( https://tcia.at/home ) website and we downloaded the results for CSCC, additionally obtained the TIDE score on the TIDE website ( http://tide.dfci.harvard.edu )[34]. In addition, alterations in genomes significantly influence the treatment response and in many instances are potent biomarkers for prediction of drug responsiveness. We obtained the information on drug sensitivity in cancer cells from The Genomics of Drug Sensitivity in Cancer (GDSC) database ( www.cancerRxgene.org ) to explore the correlation between hypoxia signature and drug responses in various cell lines Real-time quantitative polymerase chain reaction (qPCR) Total RNA was extracted from the SiHA and CasKi cells using a column-based purification method. We used ABScript III RT Master Mix for qPCR with gDNA Remover kit (ABcolony, Wuhan, China, RK20429 ) to reverse transcribe 1 µ g of total RNA to synthesize complementary DNA (cDNA) for subsequent experiments..The qPCR reaction was initiated with a master mix (ABcolonal, Wuhan, China. RK21220) containing DNA polymerase, deoxynucleotide triphosphates (dNTPs), a SYBR Green dye, template cDNA, and primers (Sangon Bioctech, Shanghai, China).We used the 2^-ΔΔCt method to relatively quantify the target gene and described the changes in gene expression levels using β - actin as a reference gene. Western Blot We used RIPA lysis buffer to lyse SiHa and CasKi cells, and measured protein concentration using BCA assay kit (Servicebio, Wuhan, China, G2026-200T). The final sample volume per well was set at 20ug. Protein was separated by 10% SDS-PAGE under constant voltage (120 V, 1.5 hours), and transferred from gel to polyvinylidene fluoride (PVDF) under 220 mA for 20 minutes. Then, the membrane was blocked with protein free rapid blocking buffer (Epizyme, Shanghai, China, PS108P) at room temperature for 1 hour, and incubated overnight with the target protein specific primary antibody (Proteintech, Wuhan, China, 13759-1-AP;20960-1-AP) at 4 ° C, followed by washing with TBST. Apply secondary antibody (Abcam, ab205718) at room temperature for 1 hour, and then wash further with TBST. Transwell migration/invasion assays For the migration assay, cells were seeded in the top chambers of Transwell plates with membrane inserts without Matrigel. For the invasion assay, the membrane inserts were precoated with an Matrigel to a uniform layer on the apical side before cells seeded.0.5 mL of DMEM supplemented with 10% FBS was added to the lower compartment of the culture plate. The plate was incubated for 24 hours to allow cell migration or invasion. Following that,We fixed the cells with 4% paraformaldehyde for 15 minutes, and then stained them with crystal violet for 30 minutes. CCK-8 assays 10 µL of CCK-8 solution (Vazyme, Nanjing, A311-01) was added to each well, and the plate was incubated for 1–4 hours. The absorbance at 450 nm was measured using a microplate reader. The optical density values obtained from the assay were used to determine the number of viable cells and assess cell proliferation or cytotoxicity. The absorbance values were subtracted by the background absorbance from blank wells and normalized to control samples to calculate the percentage of cell survival or inhibition Statistical analysis We used R 4.3.2 software for data processing, statistical analysis, and visualization, with statistical significance defined as p < 0.05. Results Identification of tumor cell state diversity The complex intra-tumoral heterogeneity poses obstacles for the diagnosis and treatment of CSCC[35–37].In order to decipher the heterogeneity in multiple samples and reveal similar or shared cellular states of cancer cells in different individuals, we integrated 5 CSCC samples from 2 publicly available cervical cancer datasets in GEO (Fig. 1 A). Top 3000 highly variable genes were remained for further analysis, followed with batch effect removal. Visualization using UMAP clustering revealed that 28537 cells from CSCC were clustered into 24 subgroups (Fig. S1 A). These clusters were labeled into different cell types based on marker genes (Fig. S1 B). We identified 11 cell types, including B cells, plasma cells, endothelial cells, epithelial cells, Macrophages, neutrophils, NK cells, fibroblasts, mast cells, and T cells (Fig. 1 B). Cancer cells were identified with inferCNV (Fig. 1 C)[38], and clustered into 7 leidens for further analysis (Fig. 1 D). To capture the intrinsic heterogeneity of cancer cells within CSCC and address their impact on patient prognosis, we applied NMF to explore consensus cellular states shared in different individuals and find cellular states that are universally and generally expressed. 8 different meta programs (MPs) were identified (Fig. 1 E). We assigned each tumor cell to the highest scoring MP based on the gene expression score of the top 30 expressed by each MP. Subsequently, we projected these cells onto the UMAP dimensionality reduction map(Fig. 1 F). To comprehensively characterize each MP and investigate their correlations with multiple aggressive tumor phenotypes, we assessed them across seven aspects including invasion, metastasis, proliferation, angiogenesis, EMT (Epithelial-Mesenchymal Transition), stemness, and hypoxia which were widely reported to related with poor outcomes of CSCC. Among them, MP7 has exhibited particularly high scores in invasion, metastasis, proliferation, hypoxia, and angiogenesis. Notably, hypoxia exlusively upregulated in mp7. These findings suggest a strong association between MP7 and hypoxia, along with a pronounced malignant phenotype, warranting the need for further in-depth exploration of MP7 (Fig. 1 G). MP7 is related with cancer aggressiveness in multiple CSCC cohorts To further reveal the biological characteristics of the hypoxia related cellular state, we identified the differentially expressed genes between MP7 and the other MP clusters (HRDEGs) (Fig. 2 A). HRDEGs with logfoldchanges > 1 and pvalue < 0.01 was included for downstream analysis. To further explore the biological functions dominanted by HRDEGs in CSCC, we revealed the related biological processes of HRDEGs through KEGG and GO enrichment analysis. In accordance with its feature of highly scored in hypoxia, our HRDEGs are mainly enriched in HIF-1 signaling pathwayas well as focal adhesion and PI3K-Akt signaling pathway (Fig. 2 B), which were reported to be involved in cell proliferation, anti-apoptosis, invasiveness, and metastasis[22, 23, 39, 40]. To validate the correlation between the aggressive tumor features presented by MP7 and the prognosis of CSCC, we conducted survival ·analysis in TCGA and GSE52903 cohort. Patients were divided into high-expression and low-expression groups based on the top 10 HRDEGs expression levels. In both datasets, the survival curve tips that high-expression of this hypoxia-related subtype predisposes patients to inferior clinical outcomes (Fig. 2 C), which suggests that MP7 may be intricately linked to tumor progression and accordingly be correlated with an unfavorable prognosis. Accordingly, we aimed to develop a signature derived from MP7 that can accurately predict the prognosis of CSCC patients, thereby facilitating early risk stratification and intervention to improve patient outcomes. To expand the sample size and enforce the re-generalizability of the prediction of this hypoxia related signature, we integrated the TCGA and GSE52903 datasets with stringent batch effect removal, enabling us to develop a more robust and comprehensive signature in a larger population with minized bias. (Fig. S1 C) Subsequently, we performed a univariate Cox regression analysis on the aforementioned HRDEGs, identifying 30 of them as significant prognostic factors for CSCC. (Fig. S1 D). In our study, we employed a rigorous selection process using 101 distinct combinations of machine learning methods to identify the optimal machine learning ensemble and the best gene combination from the aforementioned HRDEGs that are prognostic for CSCC. The signature exhibited the best C-index and contained the fewest genes in both the training and validation sets was selected. After key genes filteration by machine learning, we further refined this selection process by constructing a multivariate Cox regression model, which allowed us to identify the independent contribution of this signature on patient prognosis. The "Enet[alpha = 0.8]" combination is recognized for its fewer number of genes yet robust predictive power. (Fig. 2 D). Validation and evaluation of the hypoxia signature We identified the coefficients of the genes included in the "Enet[alpha = 0.8]" combination through multivariate Cox regression and established a hypoxia signature, which was used to calculate the risk score(HPRS) for patients and stratified them into HPRS High and HPRS Low group using the “survminer” package to identify the optimal cutoff point. Kaplan-Meier (KM) survival analysis implied that patients of HPRS high group exhibited significantly worse prognosis in comparison with those of HPRS low group. In addition, the predictive capability of HPRS was evaluated over time spans of 1, 3, and 5 years, demonstrating the area under the ROC curve (AUC) values remained stable within different duration supporting the resilience of HPRS to following periods [41] (Fig. 3 A). In the validation cohort and the independent cohort GSE52903, HPRS also showed promising prediction of shorter OS (Fig. 3 B-C). To determine whether the hypoxia signature can serve as an independent prognostic factor for CSCC patients compared to other clinical features, we conducted univariate and multivariate Cox analysis. Univariate Cox regression analysis showed that the hypoxia signature and lymph node metastasis were statistically significant related with poor outcomes, while multivariate analysis showed that they were both independent prognostic factors (Fig. 3 D). We also developed nomograms that take into account the hypoxia signature and other clinical features (Fig. 3 E). This result visually demonstrates the relationship between multiple features and the clinical outcome, showing how each variable’s value affects the predicted outcome, with the degree of impact proportional to the multivariable Cox regression coefficients. Among them, the hypoxia signature accounted for a larger proportion, intuitively demonstrating its high predictive potential. Altogether, these results confirm that the hypoxia signature is closely related to prognosis and is steady and generalizable across various patient populations. As mentioned above, we have evaluated the aggressive tumor phenotypes of multiple MPs. Coherently, these aggressive tumor phenotypes also highly correlated with the HPRS in scRNA-seq data indicating that the calculation of HPRS recapitulated its biological feature of the hypoxia signature, such as hypoxia, invasion, angiogenesis, proliferation and etc (Fig. 3 F). Furthermore, to evaluate the predictive accuracy of the HPRS and its association with the FIGO clinical staging, we expanded our dataset by including the GSE100080, GSE7410, GSE26511 and GSE146114 datasets in addition to the original training and validation datasets (TCGA and GSE52903) to increase the sample size. The analysis revealed that along with the HPRS increasing, the FIGO stage showed a significant upward trend. This is consistent with the results of the previous multivariate Cox regression analysis, which indicated that tumor advanced staging was not an independent risk factor compared to HPRS but might be consequence of high HPRS, but further verification is needed (Fig. 3 G). Analysis of tumor immune landscape between high- and low HPRS groups The complex microenvironment is a highly structured ecosystem where the cancer cells actively interact with other non-cancer components [42–44]. Hypoxia has a profound impact on the biological behavior and aggressive phenotype of cancer cells, as well as disturbation in the immune landscape in the tumor microenvironment[45–47]. It is proved that hypoxia and upregulation of hypoxia-inducible factors (HIFs) are involved in tumor immune escape and promote tumorigenesis[48, 49]. To demonstrate the distinct immune microenvironments between HPRS high and HPRS low groups, we initially used ESTIMATE to assess the stromal and immune scores of the patients. We found that the HPRS low group had a higher degree of immune score, suggesting that patients in the HPRS low group had better immune infiltration which usually denotes a better prognosis[50]. The stromal score and estimate score respectively characterize the level of infiltrating stromal cells and tumor purity in tumor tissue, and there is no significant difference between the two groups (Fig. 4 A). We further employed four different algorithms, including the MCP counter, CIBERSORT, IPS, and quanTIseq, to evaluate the distinct immune activation and related biological processes. We excluded B cells from the immune cell infiltration assessment, given the inconsistent variation of B cell infiltration deduced among different algorithms (Fig. S2 A). Consistent with previous findings, the HPRS low group showed better immune infiltration. We observed that patients in the HPRS high group were clustered and exhibited a downregulation of pro-immune features, including NK cells involved in innate immunity and DC cells involved in antigen presentation processes, CD8 T cells and activated CD4 cells involved in specific immunity were also downregulated, especially cytotoxic lymphocytes, which secrete various cytokines and play an important role in tumor killing. Meanwhile, an upregulation of immunosuppressive features, such as M2 macrophages and immune checkpoints, was also observed. On the contrary, patients in the HPRS low group were clustered into three categories, among which there were two clusters that showed upregulation of pro-immune features, while one cluster exhibited an immune phenotype similar to that of the HPRS low group.(Fig. 4 B) . Predictive value of hypoxia signature for immunotherapy Given there was a significant correlation between HPRS and immune landscape, and previous reported involvement of hypoxia in immunotherapy resistance[51], we further evaluated the efficacy of immunotherapy in different HRPS groups by ImmunoPhenoScore(IPS)and Tumor Immune Dysfunction and Exclusion༈TIDE༉ scores.By integrating diverse immunological parameters, the IPS offered a comprehensive approach to evaluating the tumor microenvironment and its potential interaction with immune checkpoint inhibitors[52].In the HPRS high group, the IPS for both single-agent PD-1 or CTLA-4 inhibitors and their combination use was significantly reduced, suggesting that higher HPRS indicates lower response level no matter using immune checkpoint inhibitors alone or in combination. This implies that the HPRS hgh group may not respond as well to immunotherapy compared to the HPRS low group (Fig. 5 A). Additionally, the TIDE score was employed to evaluate the immune evasion capacity of tumor samples by analyzing their gene expression profiles, and also serves as a predictive indicator for the efficacy of immune checkpoint inhibitors, offering valuable insights into the potential responsiveness of tumors to such therapeutic interventions. Higher TIDE scores are associated with poorer outcomes for immune checkpoint blockade therapy[53, 54].The results showed that the TIDE scores and immune exclusion scores of the HPRS high group were higher than those of the HPRS low group, while the MSI scores and dysfunction scores were lower than those of the HPRS low group, indicating a higher likelihood of immune escape (Fig. S2 B). TMB may drive a potent anti-tumor immune response, leading to a sustained clinical response to immunotherapy[55, 56].To understand the tumor mutational burden (TMB) between different HPRS groups and its correlation with HPRS, we used waterfall charts to visualize the somatic mutation maps of the HPRS High and HPRS Low groups. The total mutational burden in the HPRS high group was 81.16%, while that in the HPRS low group was 88.61%(Fig. S2 D). However, no statistically significant difference in TMB was observed between the HPRS high and HPRS low groups (Fig. S2F). The stratified survival curve analysis indicated that TMB status did not influence the risk score. Notably, the HPRS subgroups exhibited significant prognostic differences in both the low and high TMB status subgroups (Fig. 5 B). The value of hypoxia signature in chemotherapy sensitivity Hypoxia creates favorable conditions for a dominant resistance to multiple antitumor treatments, which leads to the exhibition of intrinsic or acquired resistance of chemotherapy [57–59]. We then accessed the chemo-responses and the gene expression profile of pan-cancer cell lines from GDSC[60]. The cell lines were divided into the HPRS high group and the HPRS low group as indicated above. We compared the IC50 values of common first-line chemotherapy drugs for cervical cancer between the two groups. It was observed that the IC50 of cisplatin, paclitaxel, cyclophosphamide, and gemcitabine in the HPRS low group was lower than that of the HPRS high group(Fig. 5 C). This suggests that the signature has potential predictive ability for chemotherapy drug responsiveness. P4HA2 mediates tumor hypoxia tolerance and malignant phenotype Although genes that constitute the hypoxia signature have been demonstrated to exhibit high expression levels in hypoxia-related subtypes within scRNA-seq data, it is essential to evaluate the alternation of these target genes under varying oxygen content conditions. We then assessed the expression of these genes by qPCR corroborately indicating that hypoxia induced the upregulation of these target genes in cervical cancer cell lines in vitro which supported their roles in modulating hypoxic conditions (Fig. 6 A). Among these genes, P4HA2 exhibited a significant correlation with the other five in both correlation analysis and occupied a kernel position in protein-protein interaction (PPI) networks, and it is potentially associated with tumor hypoxia (Fig. 3 F, 6 B). We assessed the impact of P4HA2 knockdown under hypoxic conditions on cell migration and invasion by plating SiHa and CaSki cells in the upper chambers of Transwell plates with and without Matrigel coating. Under the stress of hypoxia, the migration and invasion abilities of these two cell lines were significantly impaired due to P4HA2 knockdown(Fig. 6 D,E).These data indicating the biological function of P4HA2 in withstanding hypoxia is consistent with the correlation analysis results mentioned before, where P4HA2 is positively correlated with various malignant phenotype scores. Additionally, to confirm the essential role of P4HA2 in surviving hypoxic tumor conditions, we knocked down P4HA2 expression via siRNA transfection and subsequently compared the relative cell viability between untreated and P4HA2 knockdown cells after 48 hours under normoxic and hypoxic incubation. Although a decrease in cell viability under hypoxic conditions was observed in the P4HA2 knockdown group, this reduction was not pronounced. This suggested that, compared to its impact on cell survival, P4HA2 primarily affects the malignant phenotype of cells under hypoxic conditions. HIF1A plays a pivotal role in enabling cells to survive and adapt to hypoxic conditions by regulating a multitude of mechanisms, including angiogenesis, metabolic adaptation, apoptosis, and cell cycle control. As a key molecular regulator of hypoxia, HIF1A orchestrates these processes to ensure cellular homeostasis and promote cancer survival in low-oxygen environments[61, 62]. To further reveal the mechanism by which P4HA2 modulates the hypoxia tolerance and the transformation of malignant phenotypes in tumor cells. We validated the engagement of P4HA2 in regulating the hypoxia-associated pivotal molecule HIF-1α at protein level (Fig. 6 F). Under hypoxic conditions, we observed a responsive increase in both P4HA2 and HIF-1α protein levels. Additionally, after P4HA2 knockdown, we observed that HIF-1α levels decreased concomitantly with the downregulation of P4HA2, indicating an association between HIF-1α mediation and P4HA2 expression (Fig. 6 G). To further investigate whether P4HA2 affects HIF-1α at the transcriptional level, we quantified HIF1A mRNA levels following P4HA2 knockdown. However, no significant changes were detected, suggesting that P4HA2’s influence on HIF-1α via post-transcriptional modification(Fig. 6 H). Previous studies have proved that P4HA2 promotes erdafitinib resistance by regulating the stability of HIF-1α in bladder cancer with FGFR3 alteration[63]. It is widely recognized that HIF-1α undergoes hydroxylation at Pro402 and Pro564 by prolyl hydroxylases (PHDs), which facilitates its binding to von Hippel-Lindau protein (pVHL). pVHL then functions as an E3 ubiquitin ligase, promoting the ubiquitination and subsequent rapid degradation of HIF-1α via the 26S proteasome pathway. Thus, we employed the proteasome inhibitor bortezomib to block HIF-1α protein degradation[64]. After 48 hours of P4HA2 knockdown or control treatment, cells were treated with a proteasome inhibitor to inhibit the degradation of HIF-1α. Subsequently, HIF-1α protein levels were measured at 0, 1, 2, 4, and 6 hours post-inhibitor administration. In both groups, accumulation of HIF-1α was observed, indicating that the proteasome inhibitor effectively rescued the reduction in HIF-1α protein levels induced by P4HA2 knockdown (Fig. 6 I). These results suggest that P4HA2 contributes to the stabilization of HIF-1α and protects it from degradation in cervical cancer cells. Together, these data demonstrated the essensial role of P4HA2 in stablizing HIF-1α and subsequently mediating aggressive transformation while hypoxia adaptation in vitro . Discussion Compared with the favorable prognosis of localized CSCC, the 5-year survival rate for advanced, recurrent, and metastatic CSCC is only 16.5%[65]. Although radical surgery, radiochemotherapy, and their combination offer the possibility of curing early-stage and locally advanced cancers, patients with metastatic disease or those who experience persistent or recurrent cancer after chemoradiotherapy often face neglitable benifits due to non-preselection of patient potentially resistant to treatment[66]. Therefore, patient derived model untangling the heterogeneous responses are crucial to mitigate the risk of receiving ineffective therapy. However, existing stratification methods for CSCC primarily rely on multiple examination methods to identify clinical and pathological features, such as tumor stages, histological types, and lymph node metastasis which usually failed to typify the extensive intratumoral phenotypic heterogeneity and plasticity observed in cervical cancers[67, 68]. For instance, tumor mutation burden (TMB) and PD-L1 expression are widely used biomarkers for predicting immunotherapy responses, whereas neither TMB nor PD-L1 level was equivalent to the probability of treatment responses[69, 70]. Regarding this, it is demanded to investigate more precise and comprehensive stratification tools for treatment response prediction and therefore guide personalized treatment strategies. Hypoxia is a critical factor in cervical cancer progression, contributing to tumor invasion, immune evasion, and resistance to chemotherapy and radiotherapy. To date, one of the mechanisms reported to be involved in hypoxia-mediated tumor aggressiveness is about its induction of pro-angiogenic factor expression which promotes tumor growth and metastasis. Additionally, this environment pressure exerted by hypoxia could protect tumors form anti-tumor immune, and therefore reduce the efficacy of immunotherapy with reason yet unclear [58, 71–73]. For decades, hypoxia has been intensively investigated across research fields. However, hypoxia-targeted therapies are unsatisfying when translated to medical practice. One of the reasons is the lack of stratification to characterize the tumor with hypoxic features sufficiently ahead of administrating treatments due to its inherent complexity from spatial and temporal heterogeneity within the tumor microenvironment, underscoring the need for advanced methodologies [74–77]. Liu, Y., et al. utilized transcriptomic data and clinical information from gastric cancer patients in public databases to identify key genes associated with hypoxia and constructed a prognostic model[78]. Gao et al., Li et al. and other studies have also developed prognostic models related to hypoxia in other types of cancer[79–81]. Although these hypoxia-related gene signatures demonstrated promising risk stratification capability, they potentially overlook the heterogeneity of the tumor microenvironment (TME). This limitation may result in blind spots when comprehensively assessing the overall tumor immune status and hypoxic features, as the analysis of bulk RNA data could obscure critical intercellular variations in cellular composition and molecular expression patterns across distinct TME subregions[82, 83]. In contrast to previous studies, we systematically integrated bulk RNA sequencing data and scRNA-seq data to characterize the tumor cell heterogeneity in cervical squamous cell carcinoma (CSCC) patients. More recently, NMF has been applied to address gene expression commonlarity among different cells which overcomes the initial variation of either spatial or temporal conditions[84]. Utilizing NMF, we identified a hypoxia-related tumor cell subtype that strongly correlates with a multitude tumor aggressive characters like invasion, immune evasion, and resistance to chemotherapy as mentioned above. We introduced a novel hypoxia signature, HPRS, and emphasized that the genes included in this signature represent potential targets for precision therapy. Furthermore, to illustrate the underpinning mechanism of how HPRS contributes to tumor aggressiveness, we provided in vitro experimental evidence demonstrating the critical role of the key gene P4HA2 in promoting malignant phenotypes of CSCC tumor cells and its involvement in regulating HIF-1α stability. P4HA2, here in the study, identified as a key gene in our hypoxia signature, plays a crucial role in collagen synthesis and stabilization, which are essential for tumor stroma formation and invasion[85]. Validating the role of P4HA2 through experimental studies not only elucidates the underlying mechanisms of our prognostic model but also enhances its reliability and confidence. Previous research in other cancers has shown that P4HA2 is associated with poor prognosis and resistance to therapy, further supporting its potential as a biomarker in cervical cancer[63, 86–88]. However, further investigation in vivo or in preclinical models is warranted to fully understand the functional role of P4HA2 in cervical cancer and its potential as a therapeutic target. Despite the promising results, our study has several limitations. Similar to other research articles on risk stratification methods, our analysis is based on public datasets and further validation is needed in prospective studies[78–81]. The sample size is relatively small, which may limit the generalizability of our findings and the robustness of the hypoxia signature. Additionally, the study population is limited in terms of ethnic diversity, which may affect the applicability of our model to different populations. Furthermore, the inclusion of other risk factors, such as human papillomavirus (HPV) status and smoking history, could provide a more comprehensive understanding of cervical cancer prognosis. Future studies should address these limitations by incorporating larger and more diverse cohorts, validating the hypoxia signature in independent datasets, and exploring the interactions between hypoxia and other prognostic factors. In conclusion, this study delineates the complex cellular and molecular landscape of CSCC, highlights the prognostic significance of the hypoxia signature, and provides insights into the tumor immune landscape and drug responsiveness. However, further exploration may require a larger sample size to complete. The findings have implications for the development of targeted therapies and personalized treatment strategies for CSCC patients. Conclusions In this study, by combining integration of scRNA-seq and bulk RNA-seq data with multiple machine learning algorithms we establish a hypoxia associated signature as an independent prognostic factor closely related to the patient”s poor outcomes involving immune suppression profile and therapeutic resistance. To further evaluate if the hypoxia signature functionalized in modulating cervical cancer aggressive phenotypes, we identified that P4HA2 stablizes HIF-1α and accrodingly affects the hypoxia adaptation and aggressive phenotype transformation of cervical cancer cells in vitro . These data indicates that our hypoxia signature could serve as a potential therapeutic response predicator of advanced-stage CSCC patients prior to late line strategies administration. Additionally, as playing the essential in the hypoxia signature, we also demonstrate that P4HA2 should be a potential therapeutic target in treating advanced-stage CSCC patients. Abbreviations cervical squamous cell carcinoma (CSCC) Comprehensive Gene Expression Database (GEO) hypoxia-related cluster differentially expressed genes (HRDEGs) Cancer Genome Atlas (TCGA) Cervical cancer (CC) concurrent chemoradiotherapy (CCRT) Neoadjuvant chemotherapy (NACT) copy number variation (CNV) single-cell RNA sequencing (scRNA-seq) Non-negative matrix factorization (NMF) metaprograms (MPs) KEGG (Kyoto Encyclopedia of Genes and Genomes) GO (Gene Ontology) Principal Component Analysis (PCA) Super Partial Correlation (SuperPC) Random Forest, Support Vector Machine (SVM) Least Absolute Shrinkage and Selection Operator (Lasso) Gradient Boosting Machine (GBM) Kaplan-Meier (KM) tumor immune dysfunction and exclusion (TIDE) Immune phenotype score (IPS) Genomics of Drug Sensitivity in Cancer (GDSC) complementary DNA (cDNA) deoxynucleotide triphosphates (dNTPs) Epithelial-Mesenchymal Transition(EMT ) area under the ROC curve (AUC) hypoxia-related risk score（HPRS） hypoxia-inducible factors (HIFs) tumor mutational burden (TMB) prolyl hydroxylases (PHDs) protein-protein interaction (PPI) human papillomavirus (HPV) tumor microenvironment (TME) von Hippel-Lindau protein (pVHL) Declarations Ethics approval and consent to participate Not applicable Consent for publication Not applicable Availability of data and materials The data that support the findings include: scRNA-seq data (GSE208653 and GSE197461) and bulk RNA data(GSE52903, GSE100080, GSE7410, GSE26511, and GSE146114) are available from GEO (www.ncbi.nlm.nih.gov/geo/), TCGA-CESC data is available from TCGA(https://portal.gdc.cancer.gov). Data are available upon reasonable request. Competing interests The authors declare that they have no competing interests Funding This work was supported by XL Nature and Science Foundation of China (82303625) Authors' contributions Xj.L. wrote the original draft, performed Formal analysis, prepared figures and X.Z. prepared figures 1-5. Y.D. prepared figure 6. F.H. and R.W conducted data retrieval. X.H and D.M. conducted project administration. F.L and X.Li substantively revised the draft. Acknowledgements We gratefully acknowledge the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases for providing access to the scream-seq data and bulk RNA data used in this study. References Abu-Rustum NR, Yashar CM, Arend R, Barber E, Bradley K, Brooks R, Campos SM, Chino J, Chon HS, Crispens MA et al : NCCN Guidelines(R) Insights: Cervical Cancer, Version 1.2024 . J Natl Compr Canc Netw 2023, 21 (12):1224-1233. Arbyn M, Weiderpass E, Bruni L, de Sanjose S, Saraiya M, Ferlay J, Bray F: Estimates of incidence and mortality of cervical cancer in 2018: a worldwide analysis . Lancet Glob Health 2020, 8 (2):e191-e203. Zeng H, Zheng R, Sun K, Zhou M, Wang S, Li L, Chen R, Han B, Liu M, Zhou J et al : Cancer survival statistics in China 2019-2021: a multicenter, population-based study . J Natl Cancer Cent 2024, 4 (3):203-213. Lan C, Shen J, Wang Y, Li J, Liu Z, He M, Cao X, Ling J, Huang J, Zheng M et al : Camrelizumab Plus Apatinib in Patients With Advanced Cervical Cancer (CLAP): A Multicenter, Open-Label, Single-Arm, Phase II Trial . J Clin Oncol 2020, 38 (34):4095-4106. Monk BJ, Tan DSP, Hernandez Chagui JD, Takyar J, Paskow MJ, Nunes AT, Pujade-Lauraine E: Proportions and incidence of locally advanced cervical cancer: a global systematic literature review . Int J Gynecol Cancer 2022, 32 (12):1531-1539. Gadducci A, Cosio S: Neoadjuvant Chemotherapy in Locally Advanced Cervical Cancer: Review of the Literature and Perspectives of Clinical Research . Anticancer Res 2020, 40 (9):4819-4828. Cohen PA, Jhingran A, Oaknin A, Denny L: Cervical cancer . Lancet 2019, 393 (10167):169-182. Rahangdale L, Mungo C, O'Connor S, Chibwesha CJ, Brewer NT: Human papillomavirus vaccination and cervical cancer risk . BMJ 2022, 379 :e070115. Li H, Wu X, Cheng X: Advances in diagnosis and treatment of metastatic cervical cancer . J Gynecol Oncol 2016, 27 (4):e43. Delgado G, Bundy B, Zaino R, Sevin BU, Creasman WT, Major F: Prospective surgical-pathological study of disease-free interval in patients with stage IB squamous cell carcinoma of the cervix: a Gynecologic Oncology Group study . Gynecol Oncol 1990, 38 (3):352-357. Wolford JE, Tewari KS: Rational design for cervical cancer therapeutics: cellular and non-cellular based strategies on the horizon for recurrent, metastatic or refractory cervical cancer . Expert Opin Drug Discov 2018, 13 (5):445-457. Garg P, Krishna M, Subbalakshmi AR, Ramisetty S, Mohanty A, Kulkarni P, Horne D, Salgia R, Singhal SS: Emerging biomarkers and molecular targets for precision medicine in cervical cancer . Biochim Biophys Acta Rev Cancer 2024, 1879 (3):189106. Schiffman M, Castle PE, Jeronimo J, Rodriguez AC, Wacholder S: Human papillomavirus and cervical cancer . Lancet 2007, 370 (9590):890-907. Bachtiary B, Boutros PC, Pintilie M, Shi W, Bastianutto C, Li JH, Schwock J, Zhang W, Penn LZ, Jurisica I et al : Gene expression profiling in cervical cancer: an exploration of intratumor heterogeneity . Clin Cancer Res 2006, 12 (19):5632-5640. Dotto GP, Rustgi AK: Squamous Cell Cancers: A Unified Perspective on Biology and Genetics . Cancer Cell 2016, 29 (5):622-637. Rankin EB, Giaccia AJ: Hypoxic control of metastasis . Science 2016, 352 (6282):175-180. Gilkes DM, Semenza GL, Wirtz D: Hypoxia and the extracellular matrix: drivers of tumour metastasis . Nat Rev Cancer 2014, 14 (6):430-439. Horsman MR, Vaupel P: Pathophysiological Basis for the Formation of the Tumor Microenvironment . Front Oncol 2016, 6 :66. Mennerich D, Kubaichuk K, Kietzmann T: DUBs, Hypoxia, and Cancer . Trends Cancer 2019, 5 (10):632-653. Ivan M, Fishel ML, Tudoran OM, Pollok KE, Wu X, Smith PJ: Hypoxia signaling: Challenges and opportunities for cancer therapy . Semin Cancer Biol 2022, 85 :185-195. Ruan K, Song G, Ouyang G: Role of hypoxia in the hallmarks of human cancer . J Cell Biochem 2009, 107 (6):1053-1062. Semenza GL: Targeting HIF-1 for cancer therapy . Nat Rev Cancer 2003, 3 (10):721-732. Icard P, Shulman S, Farhat D, Steyaert JM, Alifano M, Lincet H: How the Warburg effect supports aggressiveness and drug resistance of cancer cells? Drug Resist Updat 2018, 38 :1-11. Taylor CT, Scholz CC: The effect of HIF on metabolism and immunity . Nat Rev Nephrol 2022, 18 (9):573-587. Guo C, Qu X, Tang X, Song Y, Wang J, Hua K, Qiu J: Spatiotemporally deciphering the mysterious mechanism of persistent HPV-induced malignant transition and immune remodelling from HPV-infected normal cervix, precancer to cervical cancer: Integrating single-cell RNA-sequencing and spatial transcriptome . Clin Transl Med 2023, 13 (3):e1219. Qu X, Wang Y, Jiang Q, Ren T, Guo C, Hua K, Qiu J: Interactions of Indoleamine 2,3-dioxygenase-expressing LAMP3(+) dendritic cells with CD4(+) regulatory T cells and CD8(+) exhausted T cells: synergistically remodeling of the immunosuppressive microenvironment in cervical cancer and therapeutic implications . Cancer Commun (Lond) 2023, 43 (11):1207-1228. Wolf FA, Angerer P, Theis FJ: SCANPY: large-scale single-cell gene expression data analysis . Genome Biol 2018, 19 (1):15. Polanski K, Young MD, Miao Z, Meyer KB, Teichmann SA, Park JE: BBKNN: fast batch alignment of single cell transcriptomes . Bioinformatics 2020, 36 (3):964-965. Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, Feng T, Zhou L, Tang W, Zhan L et al : clusterProfiler 4.0: A universal enrichment tool for interpreting omics data . Innovation (Camb) 2021, 2 (3):100141. Leek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD: The sva package for removing batch effects and other unwanted variation in high-throughput experiments . Bioinformatics 2012, 28 (6):882-883. Zhang N, Zhang H, Wu W, Zhou R, Li S, Wang Z, Dai Z, Zhang L, Liu F, Liu Z et al : Machine learning-based identification of tumor-infiltrating immune cell-associated lncRNAs for improving outcomes and immunotherapy responses in patients with low-grade glioma . Theranostics 2022, 12 (13):5931-5948. Zeng D, Ye Z, Shen R, Yu G, Wu J, Xiong Y, Zhou R, Qiu W, Huang N, Sun L et al : IOBR: Multi-Omics Immuno-Oncology Biological Research to Decode Tumor Microenvironment and Signatures . Front Immunol 2021, 12 :687975. Mayakonda A, Lin DC, Assenov Y, Plass C, Koeffler HP: Maftools: efficient and comprehensive analysis of somatic variants in cancer . Genome Res 2018, 28 (11):1747-1756. Xu Y, Wang Z, Li F: Survival prediction and response to immune checkpoint inhibitors: A prognostic immune signature for hepatocellular carcinoma . Transl Oncol 2021, 14 (1):100957. Qiu J, Qu X, Wang Y, Guo C, Lv B, Jiang Q, Su W, Wang L, Hua K: Single-Cell Landscape Highlights Heterogenous Microenvironment, Novel Immune Reaction Patterns, Potential Biomarkers and Unique Therapeutic Strategies of Cervical Squamous Carcinoma, Human Papillomavirus-Associated (HPVA) and Non-HPVA Adenocarcinoma . Adv Sci (Weinh) 2023, 10 (10):e2204951. Li C, Wu H, Guo L, Liu D, Yang S, Li S, Hua K: Single-cell transcriptomics reveals cellular heterogeneity and molecular stratification of cervical cancer . Commun Biol 2022, 5 (1):1208. Lin S, Sun Y, Cao C, Zhu Z, Xu Y, Liu B, Hu B, Peng T, Zhi W, Xu M et al : Single-nucleus RNA sequencing reveals heterogenous microenvironments and specific drug response between cervical squamous cell carcinoma and adenocarcinoma . EBioMedicine 2023, 97 :104846. Pos O, Radvanszky J, Buglyo G, Pos Z, Rusnakova D, Nagy B, Szemes T: DNA copy number variation: Main characteristics, evolutionary significance, and pathological aspects . Biomed J 2021, 44 (5):548-559. Fruman DA, Chiu H, Hopkins BD, Bagrodia S, Cantley LC, Abraham RT: The PI3K Pathway in Human Disease . Cell 2017, 170 (4):605-635. Carragher NO, Frame MC: Focal adhesion and actin dynamics: a place where kinases and proteases meet to promote invasion . Trends Cell Biol 2004, 14 (5):241-249. Metz CE: Basic principles of ROC analysis . Semin Nucl Med 1978, 8 (4):283-298. Quail DF, Joyce JA: Microenvironmental regulation of tumor progression and metastasis . Nat Med 2013, 19 (11):1423-1437. de Visser KE, Joyce JA: The evolving tumor microenvironment: From cancer initiation to metastatic outgrowth . Cancer Cell 2023, 41 (3):374-403. Goenka A, Khan F, Verma B, Sinha P, Dmello CC, Jogalekar MP, Gangadaran P, Ahn BC: Tumor microenvironment signaling and therapeutics in cancer progression . Cancer Commun (Lond) 2023, 43 (5):525-561. Wu Q, You L, Nepovimova E, Heger Z, Wu W, Kuca K, Adam V: Hypoxia-inducible factors: master regulators of hypoxic tumor immune escape . J Hematol Oncol 2022, 15 (1):77. Terry S, Buart S, Chouaib S: Hypoxic Stress-Induced Tumor and Immune Plasticity, Suppression, and Impact on Tumor Heterogeneity . Front Immunol 2017, 8 :1625. Mortezaee K, Majidpoor J: The impact of hypoxia on immune state in cancer . Life Sci 2021, 286 :120057. Vito A, El-Sayes N, Mossman K: Hypoxia-Driven Immune Escape in the Tumor Microenvironment . Cells 2020, 9 (4). Liikanen I, Lauhan C, Quon S, Omilusik K, Phan AT, Bartroli LB, Ferry A, Goulding J, Chen J, Scott-Browne JP et al : Hypoxia-inducible factor activity promotes antitumor effector function and tissue residency by CD8+ T cells . J Clin Invest 2021, 131 (7). Fridman WH, Pages F, Sautes-Fridman C, Galon J: The immune contexture in human tumours: impact on clinical outcome . Nat Rev Cancer 2012, 12 (4):298-306. Ma S, Zhao Y, Lee WC, Ong LT, Lee PL, Jiang Z, Oguz G, Niu Z, Liu M, Goh JY et al : Hypoxia induces HIF1alpha-dependent epigenetic vulnerability in triple negative breast cancer to confer immune effector dysfunction and resistance to anti-PD-1 immunotherapy . Nat Commun 2022, 13 (1):4118. Charoentong P, Finotello F, Angelova M, Mayer C, Efremova M, Rieder D, Hackl H, Trajanoski Z: Pan-cancer Immunogenomic Analyses Reveal Genotype-Immunophenotype Relationships and Predictors of Response to Checkpoint Blockade . Cell Rep 2017, 18 (1):248-262. Jiang P, Gu S, Pan D, Fu J, Sahu A, Hu X, Li Z, Traugh N, Bu X, Li B et al : Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response . Nat Med 2018, 24 (10):1550-1558. Mortezaee K, Majidpoor J: Mechanisms of CD8(+) T cell exclusion and dysfunction in cancer resistance to anti-PD-(L)1 . Biomed Pharmacother 2023, 163 :114824. Chan TA, Yarchoan M, Jaffee E, Swanton C, Quezada SA, Stenzinger A, Peters S: Development of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic . Ann Oncol 2019, 30 (1):44-56. Jardim DL, Goodman A, de Melo Gagliato D, Kurzrock R: The Challenges of Tumor Mutational Burden as an Immunotherapy Biomarker . Cancer Cell 2021, 39 (2):154-173. Teicher BA: Hypoxia and drug resistance . Cancer Metastasis Rev 1994, 13 (2):139-168. Kopecka J, Salaroglio IC, Perez-Ruiz E, Sarmento-Ribeiro AB, Saponara S, De Las Rivas J, Riganti C: Hypoxia as a driver of resistance to immunotherapy . Drug Resist Updat 2021, 59 :100787. Jing X, Yang F, Shao C, Wei K, Xie M, Shen H, Shu Y: Role of hypoxia in cancer therapy by regulating the tumor microenvironment . Mol Cancer 2019, 18 (1):157. Yang W, Soares J, Greninger P, Edelman EJ, Lightfoot H, Forbes S, Bindal N, Beare D, Smith JA, Thompson IR et al : Genomics of Drug Sensitivity in Cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells . Nucleic Acids Res 2013, 41 (Database issue):D955-961. Ravi R, Mookerjee B, Bhujwalla ZM, Sutter CH, Artemov D, Zeng Q, Dillehay LE, Madan A, Semenza GL, Bedi A: Regulation of tumor angiogenesis by p53-induced degradation of hypoxia-inducible factor 1alpha . Genes Dev 2000, 14 (1):34-44. Choudhry H, Harris AL: Advances in Hypoxia-Inducible Factor Biology . Cell Metab 2018, 27 (2):281-298. Li X, Li Y, Liu B, Chen L, Lyu F, Zhang P, He Q, Cheng L, Liu C, Song Y et al : P4HA2-mediated HIF-1alpha stabilization promotes erdafitinib-resistance in FGFR3-alteration bladder cancer . FASEB J 2023, 37 (4):e22840. Salceda S, Caro J: Hypoxia-inducible factor 1alpha (HIF-1alpha) protein is rapidly degraded by the ubiquitin-proteasome system under normoxic conditions. Its stabilization by hypoxia depends on redox-induced changes . J Biol Chem 1997, 272 (36):22642-22647. Pfaendler KS, Tewari KS: Changing paradigms in the systemic treatment of advanced cervical cancer . Am J Obstet Gynecol 2016, 214 (1):22-30. Tewari KS, Sill MW, Long HJ, 3rd, Penson RT, Huang H, Ramondetta LM, Landrum LM, Oaknin A, Reid TJ, Leitao MM et al : Improved survival with bevacizumab in advanced cervical cancer . N Engl J Med 2014, 370 (8):734-743. Nordstrom B, Strang P, Lindgren A, Bergstrom R, Tribukait B: Carcinoma of the endometrium: do the nuclear grade and DNA ploidy provide more prognostic information than do the FIGO and WHO classifications? Int J Gynecol Pathol 1996, 15 (3):191-201. Ai Y, Zhu H, Xie C, Jin X: Radiomics in cervical cancer: Current applications and future potential . Crit Rev Oncol Hematol 2020, 152 :102985. McGrail DJ, Pilie PG, Rashid NU, Voorwerk L, Slagter M, Kok M, Jonasch E, Khasraw M, Heimberger AB, Lim B et al : High tumor mutation burden fails to predict immune checkpoint blockade response across all cancer types . Ann Oncol 2021, 32 (5):661-672. Tsao MS, Le Teuff G, Shepherd FA, Landais C, Hainaut P, Filipits M, Pirker R, Le Chevalier T, Graziano S, Kratze R et al : PD-L1 protein expression assessed by immunohistochemistry is neither prognostic nor predictive of benefit from adjuvant chemotherapy in resected non-small cell lung cancer . Ann Oncol 2017, 28 (4):882-889. Graham K, Unger E: Overcoming tumor hypoxia as a barrier to radiotherapy, chemotherapy and immunotherapy in cancer treatment . Int J Nanomedicine 2018, 13 :6049-6058. Sun X, Lv X, Yan Y, Zhao Y, Ma R, He M, Wei M: Hypoxia-mediated cancer stem cell resistance and targeted therapy . Biomed Pharmacother 2020, 130 :110623. Tao J, Yang G, Zhou W, Qiu J, Chen G, Luo W, Zhao F, You L, Zheng L, Zhang T et al : Targeting hypoxic tumor microenvironment in pancreatic cancer . J Hematol Oncol 2021, 14 (1):14. Serganova I, Doubrovin M, Vider J, Ponomarev V, Soghomonyan S, Beresten T, Ageyeva L, Serganov A, Cai S, Balatoni J et al : Molecular imaging of temporal dynamics and spatial heterogeneity of hypoxia-inducible factor-1 signal transduction activity in tumors in living mice . Cancer Res 2004, 64 (17):6101-6108. Ljungkvist AS, Bussink J, Kaanders JH, Rijken PF, Begg AC, Raleigh JA, van der Kogel AJ: Hypoxic cell turnover in different solid tumor lines . Int J Radiat Oncol Biol Phys 2005, 62 (4):1157-1168. Ljungkvist AS, Bussink J, Kaanders JH, van der Kogel AJ: Dynamics of tumor hypoxia measured with bioreductive hypoxic cell markers . Radiat Res 2007, 167 (2):127-145. Cui YL, Wang X, Li XF: (18)F-fluoromisonidazole PET reveals spatial and temporal heterogeneity of hypoxia in mouse models of human non-small-cell lung cancer . Future Oncol 2015, 11 (20):2841-2849. Liu Y, Wu J, Huang W, Weng S, Wang B, Chen Y, Wang H: Development and validation of a hypoxia-immune-based microenvironment gene signature for risk stratification in gastric cancer . J Transl Med 2020, 18 (1):201. Gao F, Wang ZZ, Gu JJ, Zhang XJ, Wang HX: A Hypoxia-Associated Prognostic Gene Signature Risk Model and Prognosis Predictors in Gliomas . Frontiers in Oncology 2021, 11 . Xie SQ, Ding B, Wang SY, Zhang X, Yan WJ, Xia QQ, Meng D, Shen SY, Yu BJ, Liu HH et al : Construction of a hypoxia-immune-related prognostic model and targeted therapeutic strategies for cervical cancer . Int Immunol 2022, 34 (7):379-394. Li QX, Jin LM, Jin M: Novel Hypoxia-Related Gene Signature for Risk Stratification and Prognosis in Hepatocellular Carcinoma . Front Genet 2021, 12 . Di Giovannantonio M, Hartley F, Elshenawy B, Barberis A, Hudson D, Shafique HS, Allott VES, Harris DA, Lord SR, Haider S et al : Defining hypoxia in cancer: A landmark evaluation of hypoxia gene expression signatures . Cell Genom 2025, 5 (2):100764. Harris BH, Barberis A, West CM, Buffa FM: Gene Expression Signatures as Biomarkers of Tumour Hypoxia . Clin Oncol (R Coll Radiol) 2015, 27 (10):547-560. Rajapakse M, Tan J, Rajapakse J: Color channel encoding with NMF for face recognition . Ieee Image Proc 2004:2007-2010. Kaluz S, Zhang Q, Kuranaga Y, Yang H, Osuka S, Bhattacharya D, Devi NS, Mun J, Wang W, Zhang R et al : Targeting HIF-activated collagen prolyl 4-hydroxylase expression disrupts collagen deposition and blocks primary and metastatic uveal melanoma growth . Oncogene 2021, 40 (33):5182-5191. Zhu M, Peng R, Liang X, Lan Z, Tang M, Hou P, Song JH, Mak CSL, Park J, Zheng SE et al : P4HA2-induced prolyl hydroxylation suppresses YAP1-mediated prostate cancer cell migration, invasion, and metastasis . Oncogene 2021, 40 (41):6049-6056. Lin J, Jiang L, Wang X, Wei W, Song C, Cui Y, Wu X, Qiu G: P4HA2 Promotes Epithelial-to-Mesenchymal Transition and Glioma Malignancy through the Collagen-Dependent PI3K/AKT Pathway . J Oncol 2021, 2021 :1406853. Chi Z, Wang Q, Wang X, Li D, Tong L, Shi Y, Yang F, Guo Q, Zheng J, Chen Z: P4HA2 promotes proliferation, invasion, and metastasis through regulation of the PI3K/AKT signaling pathway in oral squamous cell carcinoma . Sci Rep 2024, 14 (1):15023. Additional Declarations No competing interests reported. Supplementary Files supplement2.pdf floatimage7.jpeg Figure S1 A Umap plot of 24 cell clusters. B Integrate bulk RNA data and remove batch effect. C Identify 14 cell types through gene labeling. D Univariate Cox regression identified HRDEGs with prognostic ability. floatimage8.jpeg Figure S2 A The heatmap illustrates the clustering patterns of B cell infiltration between HRPS high and HRPS low groups, as determined through multiple analytical approaches. B The violin chart displays the TIDE scores between HRPS high and HRPS low groups. C Waterfall charts visualize the somatic mutation maps of the HPRS High and HPRS Low groups. D Box plots compared the total mutational burden between two groups. Cite Share Download PDF Status: Published Journal Publication published 21 Jan, 2026 Read the published version in BMC Cancer → Version 1 posted Editorial decision: Revision requested 16 Oct, 2025 Reviews received at journal 13 Oct, 2025 Reviewers agreed at journal 22 Sep, 2025 Reviewers agreed at journal 22 Jul, 2025 Reviews received at journal 15 Jul, 2025 Reviewers agreed at journal 15 Jul, 2025 Reviewers invited by journal 15 Jun, 2025 Editor assigned by journal 15 Jun, 2025 Editor invited by journal 10 Jun, 2025 Submission checks completed at journal 10 Jun, 2025 First submitted to journal 10 Jun, 2025 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-6832842","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":471513487,"identity":"f740672c-c03c-4f52-af61-40015979aeea","order_by":0,"name":"Xiaojiao Li","email":"","orcid":"","institution":"Department of Gynecological Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Xiaojiao","middleName":"","lastName":"Li","suffix":""},{"id":471513488,"identity":"514f0c9b-1b18-4847-8dcf-bec89f61fa77","order_by":1,"name":"Xinyuan Zhang","email":"","orcid":"","institution":"Department of Gynecological Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Xinyuan","middleName":"","lastName":"Zhang","suffix":""},{"id":471513489,"identity":"763b9231-15c6-49dd-aaa5-92339d42d90d","order_by":2,"name":"Yilin Dai","email":"","orcid":"","institution":"Department of Gynecological Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Yilin","middleName":"","lastName":"Dai","suffix":""},{"id":471513490,"identity":"4c205995-f3c1-419f-b7e4-9515b3364770","order_by":3,"name":"Fanwei Huang","email":"","orcid":"","institution":"Department of Gynecological Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Fanwei","middleName":"","lastName":"Huang","suffix":""},{"id":471513491,"identity":"63b12577-aeec-4ec2-a5ca-1175e60f2204","order_by":4,"name":"Rui Wei","email":"","orcid":"","institution":"Department of Gynecological Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Rui","middleName":"","lastName":"Wei","suffix":""},{"id":471513492,"identity":"a310ddee-e169-4248-980d-ace8b49cbae1","order_by":5,"name":"Xiaoyuan Huang","email":"","orcid":"","institution":"Department of Gynecological Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Xiaoyuan","middleName":"","lastName":"Huang","suffix":""},{"id":471513493,"identity":"f74ce9ac-cdc7-4f13-8eca-4369939d2199","order_by":6,"name":"Ding Ma","email":"","orcid":"","institution":"Department of Gynecological Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Ding","middleName":"","lastName":"Ma","suffix":""},{"id":471513494,"identity":"014dd5d9-8704-45fe-ab25-7751af14bd3a","order_by":7,"name":"Fei Li","email":"","orcid":"","institution":"Department of Gynecological Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Fei","middleName":"","lastName":"Li","suffix":""},{"id":471513495,"identity":"ad54cd87-74a6-4d08-982f-8afe90387026","order_by":8,"name":"Xi Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3klEQVRIiWNgGAWjYDACZuaGAyCasZkZREvIEKGFseHAAbAWtgSQFh4irGFsYABbw8BjACYJajA4zth4+EPFHbvmdp7Pr27UWPAwsB8+ugGfFslmkMPOPEtubObdZp1zDOgwnrS0G/i08IP8crDtcDIjUItxDhtQiwSPGV4tbAgtPM+Mc/4RoQVmix1QC/Pj3DYitID9cubM4QRgIJsx5/ZJ8LAR8ovB+cPAAKs4bG/Yf/jx55xvdXL87IeP4dUCA4kbGxjYJMC+I0Y5CNjLAxPCB2JVj4JRMApGwcgCAGkAS4f1s2vBAAAAAElFTkSuQmCC","orcid":"","institution":"Department of Gynecological Oncology, Tongji Hospital, Tongji Medical College, Huazhong University of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"Xi","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2025-06-06 02:23:23","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6832842/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6832842/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1186/s12885-026-15597-z","type":"published","date":"2026-01-21T15:57:31+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":84792591,"identity":"4245d1b1-1665-4c8e-a021-2eb1ada8e1cc","added_by":"auto","created_at":"2025-06-17 11:40:12","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":685223,"visible":true,"origin":"","legend":"\u003cp\u003eCSCC scRNA-data data analysis displays the characteristics of tumor cell subtypes.\u003c/p\u003e\n\u003cp\u003eA Umap plot of integrated scRNA-seq profiles of 5 CSCC patients. The cell color corresponds to the sample.\u003c/p\u003e\n\u003cp\u003eB Same Umap plot as (A) but colored by cell clusters.\u003c/p\u003e\n\u003cp\u003eC Same Umap plot as (A) but colored by CNV_scores.\u003c/p\u003e\n\u003cp\u003eD Identification and clustering of NMF programs in epithelial cells with high copy number variations in CSCC patients. Different colors represent the tumor MPs identified from consensus NMF programs.\u003c/p\u003e\n\u003cp\u003eE Umap plot of epithelial cells with high copy number variations in CSCC patients, colored by cell clusters.\u003c/p\u003e\n\u003cp\u003eF Same Umap plot as (A) but colored by MPs.\u003c/p\u003e\n\u003cp\u003eG The malignant phenotype scores between MPs were displayed through dotplot and line graphs. Below the figure are tumor MPs and their clusters, while on the left are multiple malignant phenotypes to be evaluated.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/14a35ebfbc80bc768f45972b.jpeg"},{"id":84792588,"identity":"ce338a0e-7999-4546-b50a-ef9e2838f3fb","added_by":"auto","created_at":"2025-06-17 11:40:12","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":806746,"visible":true,"origin":"","legend":"\u003cp\u003eConstructing hypoxia signature based on hypoxia-related MP7\u003c/p\u003e\n\u003cp\u003eA The volcano map displays the differential genes identified between MP7 and other MPs,\u003c/p\u003e\n\u003cp\u003eB Pathway enrichment analysis of HRDEGs between MP7 and other MPs through KEGG and GO.\u003c/p\u003e\n\u003cp\u003eC Kaplan Meier analysis demonstrated the association between MP7 and OS in patients in the TCGA and GSE52903 cohorts.\u003c/p\u003e\n\u003cp\u003eD Construction of the Hypxia signature based on 101 machine learning combinations. The consistency index in the training cohort, internal validation cohort, and external validation cohort was displayed.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/37d75591ed870d1639bfe60f.jpeg"},{"id":84793000,"identity":"430fd791-8951-49fb-8c2a-a08714fd4db8","added_by":"auto","created_at":"2025-06-17 11:48:12","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":444292,"visible":true,"origin":"","legend":"\u003cp\u003eEvaluation of the prognostic potential of hypoxia signature\u003c/p\u003e\n\u003cp\u003eA-C The Kaplan–Meier survival curves demonstrated that the hypoxia signature exhibit significant prognostic potential.The ROC curves illustrated the model’s accuracy in predicting 1-year, 3-year, and 5-year survival within both the training and testing datasets.\u003c/p\u003e\n\u003cp\u003eD Perform univariate and multivariate Cox regression analysis in the TCGA cohort.\u003c/p\u003e\n\u003cp\u003eE The nomogram integrates various clinical and molecular factors to predict the 1-year, 3-year, and 5-year survival probabilities for patients with cervical squamous cell carcinoma (CSCC). Each variable is assigned a specific point value based on its influence on survival outcomes.\u003c/p\u003e\n\u003cp\u003eF The correlation heatmap shows the correlation between Hypoixa signature and various malignant phenotypes(invasion, metastasis, proliferation, hypoxia, EMT, angiogenesis).\u003c/p\u003e\n\u003cp\u003eG Box plot of HRPS in different stages (IA, IB, IIA, IIB, IIIA, IIIB, IVB). The results showed a significant correlation between staging and HRPS (Anova p=0.0054).\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/fcb8dc8ad0824afdacd7a0d3.jpeg"},{"id":84793826,"identity":"14366e6f-a7bf-4616-b13d-f2a78027f6ae","added_by":"auto","created_at":"2025-06-17 11:56:13","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":580391,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of Tumor Microenvironment\u0026nbsp; across HRPS groups.\u003c/p\u003e\n\u003cp\u003eA The box plot displays the ESTIMATE score, Immune score, and Stroma score between the HRPS\u003csup\u003eHigh\u003c/sup\u003e and\u0026nbsp; HRPS\u003csup\u003eLow\u003c/sup\u003e group. The P-value was given.\u003c/p\u003e\n\u003cp\u003eB The heatmap illustrates the distinct immune infiltration patterns between the HRPS\u003csup\u003eHigh\u003c/sup\u003e and HRPS\u003csup\u003eLow\u003c/sup\u003e groups, as characterized through comprehensive analysis using multiple computational approaches.\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/273615b7e9a601592a25aa42.jpeg"},{"id":84792593,"identity":"7e96fe4a-a4a7-45c9-b1c0-803783078b6c","added_by":"auto","created_at":"2025-06-17 11:40:13","extension":"jpeg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":398888,"visible":true,"origin":"","legend":"\u003cp\u003ePredictive value of hypoxia signature in immunotherapy.\u003c/p\u003e\n\u003cp\u003eA Viollin plot presents the differences in the IPS scores between the two risk groups stratified by CTLA4 and PD-1 treatment.\u003c/p\u003e\n\u003cp\u003eB The Kaplan–Meier curve shows the survival differences between different TMB and HPRS groups.\u003c/p\u003e\n\u003cp\u003eC Comparison of the IC50 values of chemotherapy between two risk groups, including Cisplastin, Paclitaxel, Cyclophosphamide, and Gemcitabne.\u003c/p\u003e","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/aba973adce062a6feca2730a.jpeg"},{"id":84792603,"identity":"4c1fe30f-3920-4d83-8090-910be15d59c7","added_by":"auto","created_at":"2025-06-17 11:40:13","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":497907,"visible":true,"origin":"","legend":"\u003cp\u003eVerification of the expression levels of P4HA2 and its correlation with malignant phenotype.\u003c/p\u003e\n\u003cp\u003eA The genes included in the hypoxic signature were quantified by qPCR in SiHa and CasKi cell lines under hypoxic (1% O2) and normoxic (21% O2) conditions.\u003c/p\u003e\n\u003cp\u003eB The protein-protein interaction network (PPI) showed the interaction relationship between the genes included in the hypoxic signature.\u003c/p\u003e\n\u003cp\u003eC The relative cell viability of control cells and P4HA2 knockdown cells under hypoxic conditions compared to normoxic conditions.* p \u0026lt; .05,**p \u0026lt; .01,***p \u0026lt; .001,****p \u0026lt; .0001.\u003c/p\u003e\n\u003cp\u003eD Transwell migration experiment explores the effect of P4HA2 knockdown on tumor cell invasion ability,\u003c/p\u003e\n\u003cp\u003eE Transwell invasion experiment explores the effect of P4HA2 knockdown on tumor cell metastasis ability.\u003c/p\u003e\n\u003cp\u003eF Western blot results of HIF-1α protein expression in P4HA2 knockdown and untreated groups of SiHa and CasKi cells.\u003c/p\u003e\n\u003cp\u003eG mRNA expression of HIF1A in SiHa and CasKi cells in P4HA2 knockdown and untreated groups of SiHa and CasKi cells.\u003c/p\u003e\n\u003cp\u003eH Quantitative analysis of relative expression of P4HA2 and HIF1A in Siha and CasKi cell lines after P4HA2 knockdown\u003c/p\u003e\n\u003cp\u003eI Western blot results of HIF-1α protein expression in SiHa and CasKi cells treated with proteasome inhibitors for 0.1.2.4.6 hours after P4HA2 knockdown and control treatment for 48 hours\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/da20f0cf7c8c7768771cd72c.jpeg"},{"id":101151920,"identity":"6cf081e6-b476-47dd-9b30-a1fe9adb9afb","added_by":"auto","created_at":"2026-01-26 16:08:11","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":8014227,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/6f276782-fdc5-4f22-9953-3123ba7edfd4.pdf"},{"id":84793001,"identity":"22d7d890-9f1c-41d7-8a0b-945da700a6a3","added_by":"auto","created_at":"2025-06-17 11:48:13","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":408401,"visible":true,"origin":"","legend":"","description":"","filename":"supplement2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/768a17ffeacddd45dfa8dfd8.pdf"},{"id":84793003,"identity":"f1b4eb63-aec8-40b6-9d83-773382d081b9","added_by":"auto","created_at":"2025-06-17 11:48:13","extension":"jpeg","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":558742,"visible":true,"origin":"","legend":"\u003cp\u003eFigure S1\u003c/p\u003e\n\u003cp\u003eA Umap plot of 24 cell clusters.\u003c/p\u003e\n\u003cp\u003eB Integrate bulk RNA data and remove batch effect.\u003c/p\u003e\n\u003cp\u003eC Identify 14 cell types through gene labeling.\u003c/p\u003e\n\u003cp\u003eD Univariate Cox regression identified HRDEGs with prognostic ability.\u003c/p\u003e","description":"","filename":"floatimage7.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/a6b1a72108fea2c38fc62129.jpeg"},{"id":84793004,"identity":"9030a5d9-3d98-4601-bf5c-9c5527da7dbc","added_by":"auto","created_at":"2025-06-17 11:48:13","extension":"jpeg","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":485068,"visible":true,"origin":"","legend":"\u003cp\u003eFigure S2\u003c/p\u003e\n\u003cp\u003eA The heatmap illustrates the clustering patterns of B cell infiltration between HRPS\u003csup\u003ehigh\u003c/sup\u003e and HRPS\u003csup\u003elow\u003c/sup\u003e groups, as determined through multiple analytical approaches.\u003c/p\u003e\n\u003cp\u003eB The violin chart displays the TIDE scores between HRPS\u003csup\u003ehigh\u003c/sup\u003e and HRPS\u003csup\u003elow\u003c/sup\u003e groups.\u003c/p\u003e\n\u003cp\u003eC Waterfall charts visualize the somatic mutation maps of the HPRS\u003csup\u003eHigh\u003c/sup\u003e\u0026nbsp; and HPRS\u003csup\u003eLow\u003c/sup\u003e\u0026nbsp; groups.\u003c/p\u003e\n\u003cp\u003eD Box plots compared the total mutational burden between two groups.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e","description":"","filename":"floatimage8.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6832842/v1/1f05c4632648184c977353fe.jpeg"}],"financialInterests":"No competing interests reported.","formattedTitle":"A P4HA2-dominated hypoxia signature enables response stratification of multi-therapeutic response in cervical cancer","fulltext":[{"header":"Introduction","content":"\u003cp\u003eCervical cancer (CC) was the fourth most common cancer in women world widely[1, 2]. Although the system implementation of HPV vaccines has reduced the incidence and mortality of cancer substantially, the prognosis for patients with advanced-stage, recurrent, and metastatic CC remains poor[3, 4]. Estimated proportions of 37% CC were diagnosed with locally advanced disease which were standardly treated with concurrent chemoradiotherapy (CCRT)[5], among which, around 30% of post-treatment patients would progress into recurrence or subsequent metastases[6]. Additionally, more than 10% CC cases were made up with stage IVB disease and poses an even more challenging treatment scenario[7\u0026ndash;11].Neoadjuvant chemotherapy (NACT) followed by radical hysterectomy has been considered an alternative to CCRT for stage IVB patients but with a major drawback of delaying chemo-resistant patient from on-time local treatment[6]. Early risk stratification and intervention play crucial roles in allowing patients to be treated precisely and promptly, and accordingly leading to better outcomes. A promising approach is to identify and validate novel biomarkers with prognostic and predictive significance offering insights into targetable molecular, thereby facilitating personalized cancer treatment[12].\u003c/p\u003e \u003cp\u003eCC heterogeneity is one of the major challenges of translating biomarker driven therapy into clinical practice[12]. With the advancement of recent genomic, transcriptomic and proteomic technologies, cervical squamous cell carcinoma (CSCC) is characterized with high tumor heterogeneity, accounts for approximately 80% of all CC[1, 13]. The diversity of squamous epithelial cells in CSCC encompasses of morphology, transcriptomic profiles, epigenomic modifications, metabolic activity, and accordingly treatment responses[14, 15].\u003c/p\u003e \u003cp\u003eWithin the heterogenous solid tumor microenvironment, the propensity of CSCC for hypoxia is widely revealed, which is highly involved in responsiveness to radiotherapy, chemotherapy, and immunotherapy [16\u0026ndash;18]. In contrast to normal uterine cervix tissue consuming oxygen, CSCC is depicted with hypoxia-glycolysis-acidosis paradigm due to the perturbation and deficiency in microcirculation[19]. Conversely, hypoxia itself also renders tumor heterogeneity simply by its condition of inequivalent oxygen distribution[20]. This oxygen deprivation acts indeed as a surviving stressor compelling cancer cells to adapt to this unfavorable energy condition by prompting variety of genes as p53, HIF1a or GLUT1 to evade apoptosis or necrosis [21]. Along with this adaptation, cancer cells re-allocate its energy turning out to mediate various essential processes such as proliferation, and angiogenesis, promoting tumor invasion, metastasis and treatment resistance in terms of chemotherapy, radiotherapy, and immunotherapy [22\u0026ndash;24]. HIF1a is now elucidated as an important regulator of tumor survival via manipulating hypoxic cascades. Therefore, efforts are extensively underway to decipher the role of hypoxia in CSCC to understand the tumor evolution and immune activity modulation, and consequently, improve the patient outcomes.\u003c/p\u003e \u003cp\u003eRecent studies, with the applications of high-throughput-sequencing technologies, the role of hypoxia on cancer cell heterogeneity has been elucidated in a new era with single cell resolution. Guo C. et al. presented hypoxia role in modulating macrophage polarization in 2 CCSC samples[25]. Qiu J. et al. unveiled a hypoxia related clusters as a progenitor of CC by preforming a comprehensive single-cell atlas of totally 17 samples including 3 CCSC cases[26]. Although these studies have proposed elaborate molecular mechanisms supporting tumor survival and carcinogenesis, the case sizes enrolled in these studies were usually limited with absence of follow-ups. Integration of single-cell sequencing and bulk RNA sequencing could constitute a good investigation strategy to broaden single-cell findings to larger populations. By applying 10 machine learning algorithms into 101 combinations, we visualized a significant relationship between a P4HA2 dominated hypoxia signature and poor outcomes of 297 CSCC patients \u003cem\u003ein silico\u003c/em\u003e. In this respect, we further conducted Transwell migration/invasion assays and CCK-8 assays to evaluate the role of P4HA2 with cancer cell proliferation, infiltration and aggressiveness in three cervical cell lines.\u003c/p\u003e \u003cp\u003eCollectively, herein, we identified a hypoxia signature that presents clinical significance and provided mechanical evidences of the potential to target hypoxia and P4HA2 in treating CCSC.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eData acquisition\u003c/h2\u003e \u003cp\u003eThe scRNA-seq data GSE208653 and GSE197461 are publicly available datasets from a comprehensive gene expression database(GEO, \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ncbi.nlm.nih.gov/geo/\u003c/span\u003e\u003cspan address=\"https://www.ncbi.nlm.nih.gov/geo/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). We removed samples of cervical adenocarcinoma, CIN, and normal cervical epithelium, and only retained 5 CSCC patient samples. We obtained RNA-seq data and clinical information of TCGA-CSCC (n\u0026thinsp;=\u0026thinsp;246) and GSE52903 (n\u0026thinsp;=\u0026thinsp;51), from the TCGA database and GEO dataset, respectively. In addition, datasets GSE100080, GSE7410, GSE26511, and GSE146114 were obtained from GEO to explore the correlation between risk scores and clinical staging.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003escRNAseq datasets and processing\u003c/h3\u003e\n\u003cp\u003eThe \"scanpy\" package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://scanpy.readthedocs.io/\u003c/span\u003e\u003cspan address=\"https://scanpy.readthedocs.io/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) in Python is used for quality control of scRNA seq data. We excluded cells with gene expression of less than 300 genes or more than 1000 genes, as well as cells with mitochondrial gene expression greater than 25%. We further normalized and scaled the original count, and then performed principal component analysis. [27]. Meanwhile, the \"bbknn\" package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/Teichlab/bbknn\u003c/span\u003e\u003cspan address=\"https://github.com/Teichlab/bbknn\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to remove the batch effect of scRNA-seq raw data[28]. Using unsupervised clustering analysis, we identified 24 cell clusters and displayed them using UMAP. Then, we annotated each cell cluster based on known marker genes.\u003c/p\u003e\n\u003ch3\u003eMalignant cell identification\u003c/h3\u003e\n\u003cp\u003eThe \u0026ldquo;inferCNVpy\u0026rdquo;package (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/icbi-lab/infercnvpy\u003c/span\u003e\u003cspan address=\"https://github.com/icbi-lab/infercnvpy\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was utilized in Python to infer the copy number variation (CNV) of epithelial cells in integrated single-cell RNA sequencing (scRNA-seq) data. In this analysis, T cells were used as a reference cell type to distinguish malignant epithelial cells from normal epithelial cells. Cell clusters with high CNV scores are considered malignant cells.\u003c/p\u003e\n\u003ch3\u003eExpression programs of intratumoral heterogeneity\u003c/h3\u003e\n\u003cp\u003eTo depict tumor cells heterogeneity, we performed Non-negative matrix factorization (NMF) with the \"geneNMF\" R package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/carmonalab/GeneNMF\u003c/span\u003e\u003cspan address=\"https://github.com/carmonalab/GeneNMF\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and determined the robust program with the k value was set between 3:9, and determined the consensus programs in metaprograms (MPs) that can be stably identified. In addition, We select the top 30 genes of each MP as the feature genes, and use the scanpy.tl.score_genes function to calculate the MP scores of each cell, and assign each cancer cell to the MP cluster with the highest score and project it onto the UMAP map.\u003c/p\u003e\n\u003ch3\u003eDEGs and Biological function and pathway enrichment analysis\u003c/h3\u003e\n\u003cp\u003eDEGs between MP7 and the other programs were obtained by the Wilcoxon rank-sum test, and excluded genes with logfoldchanges\u0026thinsp;\u0026lt;\u0026thinsp;1 and \u003cem\u003ep\u003c/em\u003e value\u0026thinsp;\u0026gt;\u0026thinsp;0.01. The DEGs were further used for gene set enrichment analysis through KEGG (Kyoto Encyclopedia of Genes and Genomes) and GO (Gene Ontology) with \u0026ldquo;clusterProfiler\u0026rdquo; packages(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/YuLab-SMU/clusterProfiler\u003c/span\u003e\u003cspan address=\"https://github.com/YuLab-SMU/clusterProfiler\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) in R[29]. Pathways with adjusted p-values\u0026thinsp;\u0026lt;\u0026thinsp;0.05 are considered significantly enriched in MP7\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eBatch effect removal of Bulk-RNA datasets\u003c/h2\u003e \u003cp\u003eWe Adopted the \"SVA\" package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://bioconductor.org/packages/release/bioc/html/sva.html\u003c/span\u003e\u003cspan address=\"https://bioconductor.org/packages/release/bioc/html/sva.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) to eliminate batch effects of bulk RNA data[30]. Principal Component Analysis (PCA) was performed using \"FactoMineR\"(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cran.r-project.org/web/packages/FactoMineR/index.html\u003c/span\u003e\u003cspan address=\"https://cran.r-project.org/web/packages/FactoMineR/index.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and visualized using the \"Factoextra\"(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cran.r-project.org/web//packages/factoextra/index.html\u003c/span\u003e\u003cspan address=\"https://cran.r-project.org/web//packages/factoextra/index.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) software package.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eConstruction and validation of the risk model\u003c/h3\u003e\n\u003cp\u003eThe coxph function of the \u0026ldquo;survival\u0026rdquo; package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/therneau/survival\u003c/span\u003e\u003cspan address=\"https://github.com/therneau/survival\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was used to perform univariate Cox regression analysis, the prognosis-related genes were identified based on the criteria of \u003cem\u003ep\u003c/em\u003e value\u0026thinsp;\u0026lt;\u0026thinsp;0.01. We adopted a method that encompasses 101 combinations of machine learning which has been proven to robustly identify key genes for developing reliable hypoxia signatures, including Super Partial Correlation (SuperPC), Random Forest, Support Vector Machine (SVM), Least Absolute Shrinkage and Selection Operator (Lasso), Gradient Boosting Machine (GBM), Elastic Net, Stepwise Cox, Ridge, CoxBoost, and Partial Least Squares with Cox regression (plsRcox). Among them, RSF, LASSO, CoxBoost, StepCox bidirectional, and StepCox reverse were used to perform the first step of dimensionality reduction and variable screening[31]. The TCGA data was split into a training set and an internal validation set, while GSE52903 is considered as the external validation set, and the signature with the minimum number of genes and the best C-index was ultimately selected. HRPS\u0026thinsp;=\u0026thinsp;0.9715\u0026times;EGLN1\u0026thinsp;+\u0026thinsp;0.1442\u0026times;ITGA5\u0026thinsp;\u0026minus;\u0026thinsp;0.7106\u0026times;DAPK2\u0026thinsp;+\u0026thinsp;0.2035\u0026times;PLOD2\u0026thinsp;+\u0026thinsp;0.2073\u0026times;P4HA2\u0026thinsp;+\u0026thinsp;0.6229\u0026times;AGPAT4.\u003c/p\u003e\n\u003ch3\u003eEvaluating the predictive value of the model\u003c/h3\u003e\n\u003cp\u003eWe stratified patients according to the hypoxia signature and determined the optimal cutoff value using the surv_cutpoint function in the \"survminer\" package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/kassambara/survminer\u003c/span\u003e\u003cspan address=\"https://github.com/kassambara/survminer\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Kaplan-Meier (KM) survival analysis was employed to assess the prognostic impact of the hypoxic signature. Additionally, ROC curves were utilized to validate the accuracy and stability of predicting 1-year, 3-year, and 5-year survival based on the hypoxia signature, which were generated through the \"timeROC\" package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/cran/timeROC/blob/master/R/timeROC_3.R\u003c/span\u003e\u003cspan address=\"https://github.com/cran/timeROC/blob/master/R/timeROC_3.R\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eClinical relevance of the hypoxia signature\u003c/h2\u003e \u003cp\u003eWe conducted univariate and multivariate Cox regression analysis using the \"survival\" package to determine whether the hypoxia signature is an independent prognostic factor for CSCC patients, and visualized the results using the \"forestplot\" package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cran.r-project.org/package=forestplot\u003c/span\u003e\u003cspan address=\"https://cran.r-project.org/package=forestplot\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Based on this, a nomogram of the TCGA cohort was constructed for clinical application including age, T, N, M, pathologic stage, and the hypoxia signature. Furthermore, to validate the association between the hypoxia signature and clinical staging with a larger sample size, in addition to the TCGA and GSE52903 cohorts, we included GSE100080, GSE7410, GSE146114, and GSE26511 cohorts. The relationship between Figo staging and hypoxia signature was visually demonstrated through boxplot generated by the \u0026ldquo;ggplot2\u0026rdquo; package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cran.r-project.org/package=ggplot2\u003c/span\u003e\u003cspan address=\"https://cran.r-project.org/package=ggplot2\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eImmune landscape analyses\u003c/h2\u003e \u003cp\u003eThe \"ESTIMATE\" R package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://estimate.r-forge.r-project.org/\u003c/span\u003e\u003cspan address=\"https://estimate.r-forge.r-project.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) was employed to analyze the stromal, immune, and ESTIMATE score of CSCC patients to quantify immune activation levels based on gene expression profiles. Furthermore, the assessment of the abundance of immune microenvironment and functions was conducted by the \"IBOR\" package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/IOBR/IOBR\u003c/span\u003e\u003cspan address=\"https://github.com/IOBR/IOBR\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), which includes several bioinformatics algorithms like \"MCP-counter,\" \"IPS\", \"CIBERSORT,\" and \"quanTIseq\" algorithms[32].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eTumor mutation burden and drug response analyses\u003c/h2\u003e \u003cp\u003eSNP information was collected from the TCGA database, and then the mutation profiles was analyzed based on the risk stratification using the \u0026ldquo;maftools\u0026rdquo; R package(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.bioconductor.org/packages/release/bioc/html/maftools.html\u003c/span\u003e\u003cspan address=\"https://www.bioconductor.org/packages/release/bioc/html/maftools.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e)[33]. We applied the tumor immune dysfunction and exclusion (TIDE) and the Immune phenotype score (IPS) to predict the potential immunotherapy responses in CSCC. IPS results for 20 solid tumors in TCGA are available on TCIA (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://tcia.at/home\u003c/span\u003e\u003cspan address=\"https://tcia.at/home\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) website and we downloaded the results for CSCC, additionally obtained the TIDE score on the TIDE website (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://tide.dfci.harvard.edu\u003c/span\u003e\u003cspan address=\"http://tide.dfci.harvard.edu\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e)[34]. In addition, alterations in genomes significantly influence the treatment response and in many instances are potent biomarkers for prediction of drug responsiveness. We obtained the information on drug sensitivity in cancer cells from The Genomics of Drug Sensitivity in Cancer (GDSC) database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ewww.cancerRxgene.org\u003c/a\u003e\u003c/span\u003e\u003c/span\u003e) to explore the correlation between hypoxia signature and drug responses in various cell lines\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eReal-time quantitative polymerase chain reaction (qPCR)\u003c/h2\u003e \u003cp\u003eTotal RNA was extracted from the SiHA and CasKi cells using a column-based purification method. We used ABScript III RT Master Mix for qPCR with gDNA Remover kit (ABcolony, Wuhan, China, RK20429\u003c/p\u003e \u003cp\u003e) to reverse transcribe 1 \u0026micro; g of total RNA to synthesize complementary DNA (cDNA) for subsequent experiments..The qPCR reaction was initiated with a master mix (ABcolonal, Wuhan, China. RK21220) containing DNA polymerase, deoxynucleotide triphosphates (dNTPs), a SYBR Green dye, template cDNA, and primers (Sangon Bioctech, Shanghai, China).We used the 2^-ΔΔCt method to relatively quantify the target gene and described the changes in gene expression levels using β - actin as a reference gene.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eWestern Blot\u003c/h2\u003e \u003cp\u003eWe used RIPA lysis buffer to lyse SiHa and CasKi cells, and measured protein concentration using BCA assay kit (Servicebio, Wuhan, China, G2026-200T). The final sample volume per well was set at 20ug. Protein was separated by 10% SDS-PAGE under constant voltage (120 V, 1.5 hours), and transferred from gel to polyvinylidene fluoride (PVDF) under 220 mA for 20 minutes. Then, the membrane was blocked with protein free rapid blocking buffer (Epizyme, Shanghai, China, PS108P) at room temperature for 1 hour, and incubated overnight with the target protein specific primary antibody (Proteintech, Wuhan, China, 13759-1-AP;20960-1-AP) at 4 \u0026deg; C, followed by washing with TBST. Apply secondary antibody (Abcam, ab205718) at room temperature for 1 hour, and then wash further with TBST.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eTranswell migration/invasion assays\u003c/h2\u003e \u003cp\u003eFor the migration assay, cells were seeded in the top chambers of Transwell plates with membrane inserts without Matrigel. For the invasion assay, the membrane inserts were precoated with an Matrigel to a uniform layer on the apical side before cells seeded.0.5 mL of DMEM supplemented with 10% FBS was added to the lower compartment of the culture plate. The plate was incubated for 24 hours to allow cell migration or invasion. Following that,We fixed the cells with 4% paraformaldehyde for 15 minutes, and then stained them with crystal violet for 30 minutes.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eCCK-8 assays\u003c/h2\u003e \u003cp\u003e10 \u0026micro;L of CCK-8 solution (Vazyme, Nanjing, A311-01) was added to each well, and the plate was incubated for 1\u0026ndash;4 hours. The absorbance at 450 nm was measured using a microplate reader. The optical density values obtained from the assay were used to determine the number of viable cells and assess cell proliferation or cytotoxicity. The absorbance values were subtracted by the background absorbance from blank wells and normalized to control samples to calculate the percentage of cell survival or inhibition\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eWe used R 4.3.2 software for data processing, statistical analysis, and visualization, with statistical significance defined as p\u0026thinsp;\u0026lt;\u0026thinsp;0.05.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\n \u003cdiv id=\"Sec20\" class=\"Section3\"\u003e\n \u003ch2\u003eIdentification of tumor cell state diversity\u003c/h2\u003e\n \u003cp\u003eThe complex intra-tumoral heterogeneity poses obstacles for the diagnosis and treatment of CSCC[35\u0026ndash;37].In order to decipher the heterogeneity in multiple samples and reveal similar or shared cellular states of cancer cells in different individuals, we integrated 5 CSCC samples from 2 publicly available cervical cancer datasets in GEO (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eA). Top 3000 highly variable genes were remained for further analysis, followed with batch effect removal. Visualization using UMAP clustering revealed that 28537 cells from CSCC were clustered into 24 subgroups (Fig. \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003eA). These clusters were labeled into different cell types based on marker genes (Fig. \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003eB). We identified 11 cell types, including B cells, plasma cells, endothelial cells, epithelial cells, Macrophages, neutrophils, NK cells, fibroblasts, mast cells, and T cells (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eB). Cancer cells were identified with inferCNV (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eC)[38], and clustered into 7 leidens for further analysis (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eD).\u003c/p\u003e\n \u003cp\u003eTo capture the intrinsic heterogeneity of cancer cells within CSCC and address their impact on patient prognosis, we applied NMF to explore consensus cellular states shared in different individuals and find cellular states that are universally and generally expressed. 8 different meta programs (MPs) were identified (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eE). We assigned each tumor cell to the highest scoring MP based on the gene expression score of the top 30 expressed by each MP. Subsequently, we projected these cells onto the UMAP dimensionality reduction map(Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eF).\u003c/p\u003e\n \u003cp\u003eTo comprehensively characterize each MP and investigate their correlations with multiple aggressive tumor phenotypes, we assessed them across seven aspects including invasion, metastasis, proliferation, angiogenesis, EMT (Epithelial-Mesenchymal Transition), stemness, and hypoxia which were widely reported to related with poor outcomes of CSCC. Among them, MP7 has exhibited particularly high scores in invasion, metastasis, proliferation, hypoxia, and angiogenesis. Notably, hypoxia exlusively upregulated in mp7. These findings suggest a strong association between MP7 and hypoxia, along with a pronounced malignant phenotype, warranting the need for further in-depth exploration of MP7 (Fig. \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eG).\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e\n \u003ch2\u003eMP7 is related with cancer aggressiveness in multiple CSCC cohorts\u003c/h2\u003e\n \u003cp\u003eTo further reveal the biological characteristics of the hypoxia related cellular state, we identified the differentially expressed genes between MP7 and the other MP clusters (HRDEGs) (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eA). HRDEGs with logfoldchanges\u0026thinsp;\u0026gt;\u0026thinsp;1 and \u003cem\u003epvalue\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.01 was included for downstream analysis. To further explore the biological functions dominanted by HRDEGs in CSCC, we revealed the related biological processes of HRDEGs through KEGG and GO enrichment analysis. In accordance with its feature of highly scored in hypoxia, our HRDEGs are mainly enriched in HIF-1 signaling pathwayas well as focal adhesion and PI3K-Akt signaling pathway (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eB), which were reported to be involved in cell proliferation, anti-apoptosis, invasiveness, and metastasis[22, 23, 39, 40].\u003c/p\u003e\n \u003cp\u003eTo validate the correlation between the aggressive tumor features presented by MP7 and the prognosis of CSCC, we conducted survival \u0026middot;analysis in TCGA and GSE52903 cohort. Patients were divided into high-expression and low-expression groups based on the top 10 HRDEGs expression levels. In both datasets, the survival curve tips that high-expression of this hypoxia-related subtype predisposes patients to inferior clinical outcomes (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eC), which suggests that MP7 may be intricately linked to tumor progression and accordingly be correlated with an unfavorable prognosis. Accordingly, we aimed to develop a signature derived from MP7 that can accurately predict the prognosis of CSCC patients, thereby facilitating early risk stratification and intervention to improve patient outcomes.\u003c/p\u003e\n \u003cp\u003eTo expand the sample size and enforce the re-generalizability of the prediction of this hypoxia related signature, we integrated the TCGA and GSE52903 datasets with stringent batch effect removal, enabling us to develop a more robust and comprehensive signature in a larger population with minized bias. (Fig.\u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003eC) Subsequently, we performed a univariate Cox regression analysis on the aforementioned HRDEGs, identifying 30 of them as significant prognostic factors for CSCC. (Fig.\u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003eD). In our study, we employed a rigorous selection process using 101 distinct combinations of machine learning methods to identify the optimal machine learning ensemble and the best gene combination from the aforementioned HRDEGs that are prognostic for CSCC. The signature exhibited the best C-index and contained the fewest genes in both the training and validation sets was selected. After key genes filteration by machine learning, we further refined this selection process by constructing a multivariate Cox regression model, which allowed us to identify the independent contribution of this signature on patient prognosis. The \u0026quot;Enet[alpha\u0026thinsp;=\u0026thinsp;0.8]\u0026quot; combination is recognized for its fewer number of genes yet robust predictive power. (Fig. \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eD).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\n \u003ch2\u003eValidation and evaluation of the hypoxia signature\u003c/h2\u003e\n \u003cp\u003eWe identified the coefficients of the genes included in the \u0026quot;Enet[alpha\u0026thinsp;=\u0026thinsp;0.8]\u0026quot; combination through multivariate Cox regression and established a hypoxia signature, which was used to calculate the risk score(HPRS) for patients and stratified them into HPRS\u003csup\u003eHigh\u003c/sup\u003e and HPRS\u003csup\u003eLow\u003c/sup\u003e group using the \u0026ldquo;survminer\u0026rdquo; package to identify the optimal cutoff point. Kaplan-Meier (KM) survival analysis implied that patients of HPRS\u003csup\u003ehigh\u003c/sup\u003e group exhibited significantly worse prognosis in comparison with those of HPRS\u003csup\u003elow\u003c/sup\u003e group. In addition, the predictive capability of HPRS was evaluated over time spans of 1, 3, and 5 years, demonstrating the area under the ROC curve (AUC) values remained stable within different duration supporting the resilience of HPRS to following periods [41] (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eA). In the validation cohort and the independent cohort GSE52903, HPRS also showed promising prediction of shorter OS (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eB-C). To determine whether the hypoxia signature can serve as an independent prognostic factor for CSCC patients compared to other clinical features, we conducted univariate and multivariate Cox analysis. Univariate Cox regression analysis showed that the hypoxia signature and lymph node metastasis were statistically significant related with poor outcomes, while multivariate analysis showed that they were both independent prognostic factors (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eD). We also developed nomograms that take into account the hypoxia signature and other clinical features (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eE). This result visually demonstrates the relationship between multiple features and the clinical outcome, showing how each variable\u0026rsquo;s value affects the predicted outcome, with the degree of impact proportional to the multivariable Cox regression coefficients. Among them, the hypoxia signature accounted for a larger proportion, intuitively demonstrating its high predictive potential. Altogether, these results confirm that the hypoxia signature is closely related to prognosis and is steady and generalizable across various patient populations.\u003c/p\u003e\n \u003cp\u003eAs mentioned above, we have evaluated the aggressive tumor phenotypes of multiple MPs. Coherently, these aggressive tumor phenotypes also highly correlated with the HPRS in scRNA-seq data indicating that the calculation of HPRS recapitulated its biological feature of the hypoxia signature, such as hypoxia, invasion, angiogenesis, proliferation and etc (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eF). Furthermore, to evaluate the predictive accuracy of the HPRS and its association with the FIGO clinical staging, we expanded our dataset by including the GSE100080, GSE7410, GSE26511 and GSE146114 datasets in addition to the original training and validation datasets (TCGA and GSE52903) to increase the sample size. The analysis revealed that along with the HPRS increasing, the FIGO stage showed a significant upward trend. This is consistent with the results of the previous multivariate Cox regression analysis, which indicated that tumor advanced staging was not an independent risk factor compared to HPRS but might be consequence of high HPRS, but further verification is needed (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eG).\u003c/p\u003e\n \u003cdiv id=\"Sec23\" class=\"Section3\"\u003e\n \u003ch2\u003eAnalysis of tumor immune landscape between high- and low HPRS groups\u003c/h2\u003e\n \u003cp\u003eThe complex microenvironment is a highly structured ecosystem where the cancer cells actively interact with other non-cancer components [42\u0026ndash;44]. Hypoxia has a profound impact on the biological behavior and aggressive phenotype of cancer cells, as well as disturbation in the immune landscape in the tumor microenvironment[45\u0026ndash;47]. It is proved that hypoxia and upregulation of hypoxia-inducible factors (HIFs) are involved in tumor immune escape and promote tumorigenesis[48, 49].\u003c/p\u003e\n \u003cp\u003eTo demonstrate the distinct immune microenvironments between HPRS\u003csup\u003ehigh\u003c/sup\u003e and HPRS\u003csup\u003elow\u003c/sup\u003e groups, we initially used ESTIMATE to assess the stromal and immune scores of the patients. We found that the HPRS\u003csup\u003elow\u003c/sup\u003e group had a higher degree of immune score, suggesting that patients in the HPRS\u003csup\u003elow\u003c/sup\u003e group had better immune infiltration which usually denotes a better prognosis[50]. The stromal score and estimate score respectively characterize the level of infiltrating stromal cells and tumor purity in tumor tissue, and there is no significant difference between the two groups (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eA). We further employed four different algorithms, including the MCP counter, CIBERSORT, IPS, and quanTIseq, to evaluate the distinct immune activation and related biological processes. We excluded B cells from the immune cell infiltration assessment, given the inconsistent variation of B cell infiltration deduced among different algorithms (Fig.\u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003eA). Consistent with previous findings, the HPRS\u003csup\u003elow\u003c/sup\u003e group showed better immune infiltration. We observed that patients in the HPRS\u003csup\u003ehigh\u003c/sup\u003e group were clustered and exhibited a downregulation of pro-immune features, including NK cells involved in innate immunity and DC cells involved in antigen presentation processes, CD8 T cells and activated CD4 cells involved in specific immunity were also downregulated, especially cytotoxic lymphocytes, which secrete various cytokines and play an important role in tumor killing. Meanwhile, an upregulation of immunosuppressive features, such as M2 macrophages and immune checkpoints, was also observed. On the contrary, patients in the HPRS\u003csup\u003elow\u003c/sup\u003e group were clustered into three categories, among which there were two clusters that showed upregulation of pro-immune features, while one cluster exhibited an immune phenotype similar to that of the HPRS\u003csup\u003elow\u003c/sup\u003e group.(Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003eB) .\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\n \u003ch2\u003ePredictive value of hypoxia signature for immunotherapy\u003c/h2\u003e\n \u003cp\u003eGiven there was a significant correlation between HPRS and immune landscape, and previous reported involvement of hypoxia in immunotherapy resistance[51], we further evaluated the efficacy of immunotherapy in different HRPS groups by ImmunoPhenoScore(IPS)and Tumor Immune Dysfunction and Exclusion༈TIDE༉ scores.By integrating diverse immunological parameters, the IPS offered a comprehensive approach to evaluating the tumor microenvironment and its potential interaction with immune checkpoint inhibitors[52].In the HPRS\u003csup\u003ehigh\u003c/sup\u003e group, the IPS for both single-agent PD-1 or CTLA-4 inhibitors and their combination use was significantly reduced, suggesting that higher HPRS indicates lower response level no matter using immune checkpoint inhibitors alone or in combination. This implies that the HPRS\u003csup\u003ehgh\u003c/sup\u003e group may not respond as well to immunotherapy compared to the HPRS\u003csup\u003elow\u003c/sup\u003e group (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eA). Additionally, the TIDE score was employed to evaluate the immune evasion capacity of tumor samples by analyzing their gene expression profiles, and also serves as a predictive indicator for the efficacy of immune checkpoint inhibitors, offering valuable insights into the potential responsiveness of tumors to such therapeutic interventions. Higher TIDE scores are associated with poorer outcomes for immune checkpoint blockade therapy[53, 54].The results showed that the TIDE scores and immune exclusion scores of the HPRS\u003csup\u003ehigh\u003c/sup\u003e group were higher than those of the HPRS\u003csup\u003elow\u003c/sup\u003e group, while the MSI scores and dysfunction scores were lower than those of the HPRS\u003csup\u003elow\u003c/sup\u003e group, indicating a higher likelihood of immune escape (Fig.\u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003eB).\u003c/p\u003e\n \u003cp\u003eTMB may drive a potent anti-tumor immune response, leading to a sustained clinical response to immunotherapy[55, 56].To understand the tumor mutational burden (TMB) between different HPRS groups and its correlation with HPRS, we used waterfall charts to visualize the somatic mutation maps of the HPRS\u003csup\u003eHigh\u003c/sup\u003e and HPRS\u003csup\u003eLow\u003c/sup\u003e groups. The total mutational burden in the HPRS\u003csup\u003ehigh\u003c/sup\u003e group was 81.16%, while that in the HPRS\u003csup\u003elow\u003c/sup\u003e group was 88.61%(Fig.\u003cspan class=\"InternalRef\"\u003eS2\u003c/span\u003eD). However, no statistically significant difference in TMB was observed between the HPRS\u003csup\u003ehigh\u003c/sup\u003eand HPRS\u003csup\u003elow\u003c/sup\u003e groups (Fig. S2F). The stratified survival curve analysis indicated that TMB status did not influence the risk score. Notably, the HPRS subgroups exhibited significant prognostic differences in both the low and high TMB status subgroups (Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eB).\u003c/p\u003e\n \u003cdiv id=\"Sec25\" class=\"Section3\"\u003e\n \u003ch2\u003eThe value of hypoxia signature in chemotherapy sensitivity\u003c/h2\u003e\n \u003cp\u003eHypoxia creates favorable conditions for a dominant resistance to multiple antitumor treatments, which leads to the exhibition of intrinsic or acquired resistance of chemotherapy [57\u0026ndash;59]. We then accessed the chemo-responses and the gene expression profile of pan-cancer cell lines from GDSC[60]. The cell lines were divided into the HPRS\u003csup\u003ehigh\u003c/sup\u003e group and the HPRS\u003csup\u003elow\u003c/sup\u003e group as indicated above. We compared the IC50 values of common first-line chemotherapy drugs for cervical cancer between the two groups. It was observed that the IC50 of cisplatin, paclitaxel, cyclophosphamide, and gemcitabine in the HPRS\u003csup\u003elow\u003c/sup\u003e group was lower than that of the HPRS\u003csup\u003ehigh\u003c/sup\u003e group(Fig. \u003cspan class=\"InternalRef\"\u003e5\u003c/span\u003eC). This suggests that the signature has potential predictive ability for chemotherapy drug responsiveness.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec26\" class=\"Section3\"\u003e\n \u003ch2\u003eP4HA2 mediates tumor hypoxia tolerance and malignant phenotype\u003c/h2\u003e\n \u003cp\u003eAlthough genes that constitute the hypoxia signature have been demonstrated to exhibit high expression levels in hypoxia-related subtypes within scRNA-seq data, it is essential to evaluate the alternation of these target genes under varying oxygen content conditions. We then assessed the expression of these genes by qPCR corroborately indicating that hypoxia induced the upregulation of these target genes in cervical cancer cell lines \u003cem\u003ein vitro\u003c/em\u003e which supported their roles in modulating hypoxic conditions (Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eA).\u003c/p\u003e\n \u003cp\u003eAmong these genes, P4HA2 exhibited a significant correlation with the other five in both correlation analysis and occupied a kernel position in protein-protein interaction (PPI) networks, and it is potentially associated with tumor hypoxia (Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003eF,\u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eB).\u003c/p\u003e\n \u003cp\u003eWe assessed the impact of P4HA2 knockdown under hypoxic conditions on cell migration and invasion by plating SiHa and CaSki cells in the upper chambers of Transwell plates with and without Matrigel coating. Under the stress of hypoxia, the migration and invasion abilities of these two cell lines were significantly impaired due to P4HA2 knockdown(Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eD,E).These data indicating the biological function of P4HA2 in withstanding hypoxia is consistent with the correlation analysis results mentioned before, where P4HA2 is positively correlated with various malignant phenotype scores. Additionally, to confirm the essential role of P4HA2 in surviving hypoxic tumor conditions, we knocked down P4HA2 expression via siRNA transfection and subsequently compared the relative cell viability between untreated and P4HA2 knockdown cells after 48 hours under normoxic and hypoxic incubation. Although a decrease in cell viability under hypoxic conditions was observed in the P4HA2 knockdown group, this reduction was not pronounced. This suggested that, compared to its impact on cell survival, P4HA2 primarily affects the malignant phenotype of cells under hypoxic conditions.\u003c/p\u003e\n \u003cp\u003eHIF1A plays a pivotal role in enabling cells to survive and adapt to hypoxic conditions by regulating a multitude of mechanisms, including angiogenesis, metabolic adaptation, apoptosis, and cell cycle control. As a key molecular regulator of hypoxia, HIF1A orchestrates these processes to ensure cellular homeostasis and promote cancer survival in low-oxygen environments[61, 62]. To further reveal the mechanism by which P4HA2 modulates the hypoxia tolerance and the transformation of malignant phenotypes in tumor cells. We validated the engagement of P4HA2 in regulating the hypoxia-associated pivotal molecule HIF-1\u0026alpha; at protein level (Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eF). Under hypoxic conditions, we observed a responsive increase in both P4HA2 and HIF-1\u0026alpha; protein levels. Additionally, after P4HA2 knockdown, we observed that HIF-1\u0026alpha; levels decreased concomitantly with the downregulation of P4HA2, indicating an association between HIF-1\u0026alpha; mediation and P4HA2 expression (Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eG). To further investigate whether P4HA2 affects HIF-1\u0026alpha; at the transcriptional level, we quantified HIF1A mRNA levels following P4HA2 knockdown. However, no significant changes were detected, suggesting that P4HA2\u0026rsquo;s influence on HIF-1\u0026alpha; via post-transcriptional modification(Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eH). Previous studies have proved that P4HA2 promotes erdafitinib resistance by regulating the stability of HIF-1\u0026alpha; in bladder cancer with FGFR3 alteration[63]. It is widely recognized that HIF-1\u0026alpha; undergoes hydroxylation at Pro402 and Pro564 by prolyl hydroxylases (PHDs), which facilitates its binding to von Hippel-Lindau protein (pVHL). pVHL then functions as an E3 ubiquitin ligase, promoting the ubiquitination and subsequent rapid degradation of HIF-1\u0026alpha; via the 26S proteasome pathway. Thus, we employed the proteasome inhibitor bortezomib to block HIF-1\u0026alpha; protein degradation[64]. After 48 hours of P4HA2 knockdown or control treatment, cells were treated with a proteasome inhibitor to inhibit the degradation of HIF-1\u0026alpha;. Subsequently, HIF-1\u0026alpha; protein levels were measured at 0, 1, 2, 4, and 6 hours post-inhibitor administration. In both groups, accumulation of HIF-1\u0026alpha; was observed, indicating that the proteasome inhibitor effectively rescued the reduction in HIF-1\u0026alpha; protein levels induced by P4HA2 knockdown (Fig. \u003cspan class=\"InternalRef\"\u003e6\u003c/span\u003eI). These results suggest that P4HA2 contributes to the stabilization of HIF-1\u0026alpha; and protects it from degradation in cervical cancer cells.\u003c/p\u003e\n \u003cp\u003eTogether, these data demonstrated the essensial role of P4HA2 in stablizing HIF-1\u0026alpha; and subsequently mediating aggressive transformation while hypoxia adaptation \u003cem\u003ein vitro\u003c/em\u003e.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eCompared with the favorable prognosis of localized CSCC, the 5-year survival rate for advanced, recurrent, and metastatic CSCC is only 16.5%[65]. Although radical surgery, radiochemotherapy, and their combination offer the possibility of curing early-stage and locally advanced cancers, patients with metastatic disease or those who experience persistent or recurrent cancer after chemoradiotherapy often face neglitable benifits due to non-preselection of patient potentially resistant to treatment[66]. Therefore, patient derived model untangling the heterogeneous responses are crucial to mitigate the risk of receiving ineffective therapy. However, existing stratification methods for CSCC primarily rely on multiple examination methods to identify clinical and pathological features, such as tumor stages, histological types, and lymph node metastasis which usually failed to typify the extensive intratumoral phenotypic heterogeneity and plasticity observed in cervical cancers[67, 68]. For instance, tumor mutation burden (TMB) and PD-L1 expression are widely used biomarkers for predicting immunotherapy responses, whereas neither TMB nor PD-L1 level was equivalent to the probability of treatment responses[69, 70]. Regarding this, it is demanded to investigate more precise and comprehensive stratification tools for treatment response prediction and therefore guide personalized treatment strategies.\u003c/p\u003e \u003cp\u003eHypoxia is a critical factor in cervical cancer progression, contributing to tumor invasion, immune evasion, and resistance to chemotherapy and radiotherapy. To date, one of the mechanisms reported to be involved in hypoxia-mediated tumor aggressiveness is about its induction of pro-angiogenic factor expression which promotes tumor growth and metastasis. Additionally, this environment pressure exerted by hypoxia could protect tumors form anti-tumor immune, and therefore reduce the efficacy of immunotherapy with reason yet unclear [58, 71\u0026ndash;73]. For decades, hypoxia has been intensively investigated across research fields. However, hypoxia-targeted therapies are unsatisfying when translated to medical practice. One of the reasons is the lack of stratification to characterize the tumor with hypoxic features sufficiently ahead of administrating treatments due to its inherent complexity from spatial and temporal heterogeneity within the tumor microenvironment, underscoring the need for advanced methodologies [74\u0026ndash;77]. Liu, Y., et al. utilized transcriptomic data and clinical information from gastric cancer patients in public databases to identify key genes associated with hypoxia and constructed a prognostic model[78]. Gao et al., Li et al. and other studies have also developed prognostic models related to hypoxia in other types of cancer[79\u0026ndash;81]. Although these hypoxia-related gene signatures demonstrated promising risk stratification capability, they potentially overlook the heterogeneity of the tumor microenvironment (TME). This limitation may result in blind spots when comprehensively assessing the overall tumor immune status and hypoxic features, as the analysis of bulk RNA data could obscure critical intercellular variations in cellular composition and molecular expression patterns across distinct TME subregions[82, 83]. In contrast to previous studies, we systematically integrated bulk RNA sequencing data and scRNA-seq data to characterize the tumor cell heterogeneity in cervical squamous cell carcinoma (CSCC) patients. More recently, NMF has been applied to address gene expression commonlarity among different cells which overcomes the initial variation of either spatial or temporal conditions[84]. Utilizing NMF, we identified a hypoxia-related tumor cell subtype that strongly correlates with a multitude tumor aggressive characters like invasion, immune evasion, and resistance to chemotherapy as mentioned above. We introduced a novel hypoxia signature, HPRS, and emphasized that the genes included in this signature represent potential targets for precision therapy. Furthermore, to illustrate the underpinning mechanism of how HPRS contributes to tumor aggressiveness, we provided \u003cem\u003ein vitro\u003c/em\u003e experimental evidence demonstrating the critical role of the key gene P4HA2 in promoting malignant phenotypes of CSCC tumor cells and its involvement in regulating HIF-1α stability.\u003c/p\u003e \u003cp\u003eP4HA2, here in the study, identified as a key gene in our hypoxia signature, plays a crucial role in collagen synthesis and stabilization, which are essential for tumor stroma formation and invasion[85]. Validating the role of P4HA2 through experimental studies not only elucidates the underlying mechanisms of our prognostic model but also enhances its reliability and confidence. Previous research in other cancers has shown that P4HA2 is associated with poor prognosis and resistance to therapy, further supporting its potential as a biomarker in cervical cancer[63, 86\u0026ndash;88]. However, further investigation \u003cem\u003ein vivo\u003c/em\u003e or in preclinical models is warranted to fully understand the functional role of P4HA2 in cervical cancer and its potential as a therapeutic target.\u003c/p\u003e \u003cp\u003eDespite the promising results, our study has several limitations. Similar to other research articles on risk stratification methods, our analysis is based on public datasets and further validation is needed in prospective studies[78\u0026ndash;81]. The sample size is relatively small, which may limit the generalizability of our findings and the robustness of the hypoxia signature. Additionally, the study population is limited in terms of ethnic diversity, which may affect the applicability of our model to different populations. Furthermore, the inclusion of other risk factors, such as human papillomavirus (HPV) status and smoking history, could provide a more comprehensive understanding of cervical cancer prognosis. Future studies should address these limitations by incorporating larger and more diverse cohorts, validating the hypoxia signature in independent datasets, and exploring the interactions between hypoxia and other prognostic factors.\u003c/p\u003e \u003cp\u003eIn conclusion, this study delineates the complex cellular and molecular landscape of CSCC, highlights the prognostic significance of the hypoxia signature, and provides insights into the tumor immune landscape and drug responsiveness. However, further exploration may require a larger sample size to complete. The findings have implications for the development of targeted therapies and personalized treatment strategies for CSCC patients.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eIn this study, by combining integration of scRNA-seq and bulk RNA-seq data with multiple machine learning algorithms we establish a hypoxia associated signature as an independent prognostic factor closely related to the patient\u0026rdquo;s poor outcomes involving immune suppression profile and therapeutic resistance. To further evaluate if the hypoxia signature functionalized in modulating cervical cancer aggressive phenotypes, we identified that P4HA2 stablizes HIF-1α and accrodingly affects the hypoxia adaptation and aggressive phenotype transformation of cervical cancer cells \u003cem\u003ein vitro\u003c/em\u003e. These data indicates that our hypoxia signature could serve as a potential therapeutic response predicator of advanced-stage CSCC patients prior to late line strategies administration. Additionally, as playing the essential in the hypoxia signature, we also demonstrate that P4HA2 should be a potential therapeutic target in treating advanced-stage CSCC patients.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003ecervical squamous cell carcinoma (CSCC)\u003c/p\u003e\n\u003cp\u003eComprehensive Gene Expression Database (GEO)\u003c/p\u003e\n\u003cp\u003ehypoxia-related cluster differentially expressed genes (HRDEGs)\u003c/p\u003e\n\u003cp\u003eCancer Genome Atlas (TCGA)\u003c/p\u003e\n\u003cp\u003eCervical cancer (CC)\u003c/p\u003e\n\u003cp\u003econcurrent chemoradiotherapy (CCRT)\u003c/p\u003e\n\u003cp\u003eNeoadjuvant chemotherapy (NACT)\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;copy number variation (CNV)\u003c/p\u003e\n\u003cp\u003esingle-cell RNA sequencing (scRNA-seq)\u003c/p\u003e\n\u003cp\u003eNon-negative matrix factorization (NMF)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003emetaprograms (MPs)\u003c/p\u003e\n\u003cp\u003eKEGG (Kyoto Encyclopedia of Genes and Genomes)\u003c/p\u003e\n\u003cp\u003eGO (Gene Ontology)\u003c/p\u003e\n\u003cp\u003ePrincipal Component Analysis (PCA)\u003c/p\u003e\n\u003cp\u003eSuper Partial Correlation (SuperPC)\u003c/p\u003e\n\u003cp\u003eRandom Forest, Support Vector Machine (SVM)\u003c/p\u003e\n\u003cp\u003eLeast Absolute Shrinkage and Selection Operator (Lasso)\u003c/p\u003e\n\u003cp\u003eGradient Boosting Machine (GBM)\u003c/p\u003e\n\u003cp\u003eKaplan-Meier (KM)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003etumor immune dysfunction and exclusion (TIDE) \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Immune phenotype score (IPS)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eGenomics of Drug Sensitivity in Cancer (GDSC)\u003c/p\u003e\n\u003cp\u003ecomplementary DNA (cDNA)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003edeoxynucleotide triphosphates (dNTPs)\u003c/p\u003e\n\u003cp\u003eEpithelial-Mesenchymal Transition(EMT )\u003c/p\u003e\n\u003cp\u003earea under the ROC curve (AUC)\u003c/p\u003e\n\u003cp\u003ehypoxia-related risk score（HPRS）\u003c/p\u003e\n\u003cp\u003ehypoxia-inducible factors (HIFs)\u003c/p\u003e\n\u003cp\u003etumor mutational burden (TMB)\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;prolyl hydroxylases (PHDs)\u003c/p\u003e\n\u003cp\u003eprotein-protein interaction (PPI)\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ehuman papillomavirus (HPV)\u003c/p\u003e\n\u003cp\u003etumor microenvironment (TME)\u003c/p\u003e\n\u003cp\u003evon Hippel-Lindau protein (pVHL)\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data that support the findings include: scRNA-seq data (GSE208653 and GSE197461) and bulk RNA data(GSE52903, GSE100080, GSE7410, GSE26511, and GSE146114) are available from GEO (www.ncbi.nlm.nih.gov/geo/), TCGA-CESC data is available \u0026nbsp;from TCGA(https://portal.gdc.cancer.gov). Data are available upon reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by XL Nature and Science Foundation of China (82303625)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthors' contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eXj.L. wrote the original draft, performed Formal analysis, prepared figures and X.Z. \u0026nbsp;prepared figures 1-5. Y.D. prepared figure 6. F.H. and R.W conducted data retrieval. X.H and D.M. conducted project administration. F.L and X.Li substantively revised the draft.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe gratefully acknowledge the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA) databases for providing access to the scream-seq data and bulk RNA data used in this study.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAbu-Rustum NR, Yashar CM, Arend R, Barber E, Bradley K, Brooks R, Campos SM, Chino J, Chon HS, Crispens MA\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eNCCN Guidelines(R) Insights: Cervical Cancer, Version 1.2024\u003c/strong\u003e. \u003cem\u003eJ Natl Compr Canc Netw \u003c/em\u003e2023, \u003cstrong\u003e21\u003c/strong\u003e(12):1224-1233.\u003c/li\u003e\n\u003cli\u003eArbyn M, Weiderpass E, Bruni L, de Sanjose S, Saraiya M, Ferlay J, Bray F: \u003cstrong\u003eEstimates of incidence and mortality of cervical cancer in 2018: a worldwide analysis\u003c/strong\u003e. \u003cem\u003eLancet Glob Health \u003c/em\u003e2020, \u003cstrong\u003e8\u003c/strong\u003e(2):e191-e203.\u003c/li\u003e\n\u003cli\u003eZeng H, Zheng R, Sun K, Zhou M, Wang S, Li L, Chen R, Han B, Liu M, Zhou J\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eCancer survival statistics in China 2019-2021: a multicenter, population-based study\u003c/strong\u003e. \u003cem\u003eJ Natl Cancer Cent \u003c/em\u003e2024, \u003cstrong\u003e4\u003c/strong\u003e(3):203-213.\u003c/li\u003e\n\u003cli\u003eLan C, Shen J, Wang Y, Li J, Liu Z, He M, Cao X, Ling J, Huang J, Zheng M\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eCamrelizumab Plus Apatinib in Patients With Advanced Cervical Cancer (CLAP): A Multicenter, Open-Label, Single-Arm, Phase II Trial\u003c/strong\u003e. \u003cem\u003eJ Clin Oncol \u003c/em\u003e2020, \u003cstrong\u003e38\u003c/strong\u003e(34):4095-4106.\u003c/li\u003e\n\u003cli\u003eMonk BJ, Tan DSP, Hernandez Chagui JD, Takyar J, Paskow MJ, Nunes AT, Pujade-Lauraine E: \u003cstrong\u003eProportions and incidence of locally advanced cervical cancer: a global systematic literature review\u003c/strong\u003e. \u003cem\u003eInt J Gynecol Cancer \u003c/em\u003e2022, \u003cstrong\u003e32\u003c/strong\u003e(12):1531-1539.\u003c/li\u003e\n\u003cli\u003eGadducci A, Cosio S: \u003cstrong\u003eNeoadjuvant Chemotherapy in Locally Advanced Cervical Cancer: Review of the Literature and Perspectives of Clinical Research\u003c/strong\u003e. \u003cem\u003eAnticancer Res \u003c/em\u003e2020, \u003cstrong\u003e40\u003c/strong\u003e(9):4819-4828.\u003c/li\u003e\n\u003cli\u003eCohen PA, Jhingran A, Oaknin A, Denny L: \u003cstrong\u003eCervical cancer\u003c/strong\u003e. \u003cem\u003eLancet \u003c/em\u003e2019, \u003cstrong\u003e393\u003c/strong\u003e(10167):169-182.\u003c/li\u003e\n\u003cli\u003eRahangdale L, Mungo C, O'Connor S, Chibwesha CJ, Brewer NT: \u003cstrong\u003eHuman papillomavirus vaccination and cervical cancer risk\u003c/strong\u003e. \u003cem\u003eBMJ \u003c/em\u003e2022, \u003cstrong\u003e379\u003c/strong\u003e:e070115.\u003c/li\u003e\n\u003cli\u003eLi H, Wu X, Cheng X: \u003cstrong\u003eAdvances in diagnosis and treatment of metastatic cervical cancer\u003c/strong\u003e. \u003cem\u003eJ Gynecol Oncol \u003c/em\u003e2016, \u003cstrong\u003e27\u003c/strong\u003e(4):e43.\u003c/li\u003e\n\u003cli\u003eDelgado G, Bundy B, Zaino R, Sevin BU, Creasman WT, Major F: \u003cstrong\u003eProspective surgical-pathological study of disease-free interval in patients with stage IB squamous cell carcinoma of the cervix: a Gynecologic Oncology Group study\u003c/strong\u003e. \u003cem\u003eGynecol Oncol \u003c/em\u003e1990, \u003cstrong\u003e38\u003c/strong\u003e(3):352-357.\u003c/li\u003e\n\u003cli\u003eWolford JE, Tewari KS: \u003cstrong\u003eRational design for cervical cancer therapeutics: cellular and non-cellular based strategies on the horizon for recurrent, metastatic or refractory cervical cancer\u003c/strong\u003e. \u003cem\u003eExpert Opin Drug Discov \u003c/em\u003e2018, \u003cstrong\u003e13\u003c/strong\u003e(5):445-457.\u003c/li\u003e\n\u003cli\u003eGarg P, Krishna M, Subbalakshmi AR, Ramisetty S, Mohanty A, Kulkarni P, Horne D, Salgia R, Singhal SS: \u003cstrong\u003eEmerging biomarkers and molecular targets for precision medicine in cervical cancer\u003c/strong\u003e. \u003cem\u003eBiochim Biophys Acta Rev Cancer \u003c/em\u003e2024, \u003cstrong\u003e1879\u003c/strong\u003e(3):189106.\u003c/li\u003e\n\u003cli\u003eSchiffman M, Castle PE, Jeronimo J, Rodriguez AC, Wacholder S: \u003cstrong\u003eHuman papillomavirus and cervical cancer\u003c/strong\u003e. \u003cem\u003eLancet \u003c/em\u003e2007, \u003cstrong\u003e370\u003c/strong\u003e(9590):890-907.\u003c/li\u003e\n\u003cli\u003eBachtiary B, Boutros PC, Pintilie M, Shi W, Bastianutto C, Li JH, Schwock J, Zhang W, Penn LZ, Jurisica I\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eGene expression profiling in cervical cancer: an exploration of intratumor heterogeneity\u003c/strong\u003e. \u003cem\u003eClin Cancer Res \u003c/em\u003e2006, \u003cstrong\u003e12\u003c/strong\u003e(19):5632-5640.\u003c/li\u003e\n\u003cli\u003eDotto GP, Rustgi AK: \u003cstrong\u003eSquamous Cell Cancers: A Unified Perspective on Biology and Genetics\u003c/strong\u003e. \u003cem\u003eCancer Cell \u003c/em\u003e2016, \u003cstrong\u003e29\u003c/strong\u003e(5):622-637.\u003c/li\u003e\n\u003cli\u003eRankin EB, Giaccia AJ: \u003cstrong\u003eHypoxic control of metastasis\u003c/strong\u003e. \u003cem\u003eScience \u003c/em\u003e2016, \u003cstrong\u003e352\u003c/strong\u003e(6282):175-180.\u003c/li\u003e\n\u003cli\u003eGilkes DM, Semenza GL, Wirtz D: \u003cstrong\u003eHypoxia and the extracellular matrix: drivers of tumour metastasis\u003c/strong\u003e. \u003cem\u003eNat Rev Cancer \u003c/em\u003e2014, \u003cstrong\u003e14\u003c/strong\u003e(6):430-439.\u003c/li\u003e\n\u003cli\u003eHorsman MR, Vaupel P: \u003cstrong\u003ePathophysiological Basis for the Formation of the Tumor Microenvironment\u003c/strong\u003e. \u003cem\u003eFront Oncol \u003c/em\u003e2016, \u003cstrong\u003e6\u003c/strong\u003e:66.\u003c/li\u003e\n\u003cli\u003eMennerich D, Kubaichuk K, Kietzmann T: \u003cstrong\u003eDUBs, Hypoxia, and Cancer\u003c/strong\u003e. \u003cem\u003eTrends Cancer \u003c/em\u003e2019, \u003cstrong\u003e5\u003c/strong\u003e(10):632-653.\u003c/li\u003e\n\u003cli\u003eIvan M, Fishel ML, Tudoran OM, Pollok KE, Wu X, Smith PJ: \u003cstrong\u003eHypoxia signaling: Challenges and opportunities for cancer therapy\u003c/strong\u003e. \u003cem\u003eSemin Cancer Biol \u003c/em\u003e2022, \u003cstrong\u003e85\u003c/strong\u003e:185-195.\u003c/li\u003e\n\u003cli\u003eRuan K, Song G, Ouyang G: \u003cstrong\u003eRole of hypoxia in the hallmarks of human cancer\u003c/strong\u003e. \u003cem\u003eJ Cell Biochem \u003c/em\u003e2009, \u003cstrong\u003e107\u003c/strong\u003e(6):1053-1062.\u003c/li\u003e\n\u003cli\u003eSemenza GL: \u003cstrong\u003eTargeting HIF-1 for cancer therapy\u003c/strong\u003e. \u003cem\u003eNat Rev Cancer \u003c/em\u003e2003, \u003cstrong\u003e3\u003c/strong\u003e(10):721-732.\u003c/li\u003e\n\u003cli\u003eIcard P, Shulman S, Farhat D, Steyaert JM, Alifano M, Lincet H: \u003cstrong\u003eHow the Warburg effect supports aggressiveness and drug resistance of cancer cells?\u003c/strong\u003e \u003cem\u003eDrug Resist Updat \u003c/em\u003e2018, \u003cstrong\u003e38\u003c/strong\u003e:1-11.\u003c/li\u003e\n\u003cli\u003eTaylor CT, Scholz CC: \u003cstrong\u003eThe effect of HIF on metabolism and immunity\u003c/strong\u003e. \u003cem\u003eNat Rev Nephrol \u003c/em\u003e2022, \u003cstrong\u003e18\u003c/strong\u003e(9):573-587.\u003c/li\u003e\n\u003cli\u003eGuo C, Qu X, Tang X, Song Y, Wang J, Hua K, Qiu J: \u003cstrong\u003eSpatiotemporally deciphering the mysterious mechanism of persistent HPV-induced malignant transition and immune remodelling from HPV-infected normal cervix, precancer to cervical cancer: Integrating single-cell RNA-sequencing and spatial transcriptome\u003c/strong\u003e. \u003cem\u003eClin Transl Med \u003c/em\u003e2023, \u003cstrong\u003e13\u003c/strong\u003e(3):e1219.\u003c/li\u003e\n\u003cli\u003eQu X, Wang Y, Jiang Q, Ren T, Guo C, Hua K, Qiu J: \u003cstrong\u003eInteractions of Indoleamine 2,3-dioxygenase-expressing LAMP3(+) dendritic cells with CD4(+) regulatory T cells and CD8(+) exhausted T cells: synergistically remodeling of the immunosuppressive microenvironment in cervical cancer and therapeutic implications\u003c/strong\u003e. \u003cem\u003eCancer Commun (Lond) \u003c/em\u003e2023, \u003cstrong\u003e43\u003c/strong\u003e(11):1207-1228.\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 Biol \u003c/em\u003e2018, \u003cstrong\u003e19\u003c/strong\u003e(1):15.\u003c/li\u003e\n\u003cli\u003ePolanski K, Young MD, Miao Z, Meyer KB, Teichmann SA, Park JE: \u003cstrong\u003eBBKNN: fast batch alignment of single cell transcriptomes\u003c/strong\u003e. \u003cem\u003eBioinformatics \u003c/em\u003e2020, \u003cstrong\u003e36\u003c/strong\u003e(3):964-965.\u003c/li\u003e\n\u003cli\u003eWu T, Hu E, Xu S, Chen M, Guo P, Dai Z, Feng T, Zhou L, Tang W, Zhan L\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eclusterProfiler 4.0: A universal enrichment tool for interpreting omics data\u003c/strong\u003e. \u003cem\u003eInnovation (Camb) \u003c/em\u003e2021, \u003cstrong\u003e2\u003c/strong\u003e(3):100141.\u003c/li\u003e\n\u003cli\u003eLeek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD: \u003cstrong\u003eThe sva package for removing batch effects and other unwanted variation in high-throughput experiments\u003c/strong\u003e. \u003cem\u003eBioinformatics \u003c/em\u003e2012, \u003cstrong\u003e28\u003c/strong\u003e(6):882-883.\u003c/li\u003e\n\u003cli\u003eZhang N, Zhang H, Wu W, Zhou R, Li S, Wang Z, Dai Z, Zhang L, Liu F, Liu Z\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eMachine learning-based identification of tumor-infiltrating immune cell-associated lncRNAs for improving outcomes and immunotherapy responses in patients with low-grade glioma\u003c/strong\u003e. \u003cem\u003eTheranostics \u003c/em\u003e2022, \u003cstrong\u003e12\u003c/strong\u003e(13):5931-5948.\u003c/li\u003e\n\u003cli\u003eZeng D, Ye Z, Shen R, Yu G, Wu J, Xiong Y, Zhou R, Qiu W, Huang N, Sun L\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eIOBR: Multi-Omics Immuno-Oncology Biological Research to Decode Tumor Microenvironment and Signatures\u003c/strong\u003e. \u003cem\u003eFront Immunol \u003c/em\u003e2021, \u003cstrong\u003e12\u003c/strong\u003e:687975.\u003c/li\u003e\n\u003cli\u003eMayakonda A, Lin DC, Assenov Y, Plass C, Koeffler HP: \u003cstrong\u003eMaftools: efficient and comprehensive analysis of somatic variants in cancer\u003c/strong\u003e. \u003cem\u003eGenome Res \u003c/em\u003e2018, \u003cstrong\u003e28\u003c/strong\u003e(11):1747-1756.\u003c/li\u003e\n\u003cli\u003eXu Y, Wang Z, Li F: \u003cstrong\u003eSurvival prediction and response to immune checkpoint inhibitors: A prognostic immune signature for hepatocellular carcinoma\u003c/strong\u003e. \u003cem\u003eTransl Oncol \u003c/em\u003e2021, \u003cstrong\u003e14\u003c/strong\u003e(1):100957.\u003c/li\u003e\n\u003cli\u003eQiu J, Qu X, Wang Y, Guo C, Lv B, Jiang Q, Su W, Wang L, Hua K: \u003cstrong\u003eSingle-Cell Landscape Highlights Heterogenous Microenvironment, Novel Immune Reaction Patterns, Potential Biomarkers and Unique Therapeutic Strategies of Cervical Squamous Carcinoma, Human Papillomavirus-Associated (HPVA) and Non-HPVA Adenocarcinoma\u003c/strong\u003e. \u003cem\u003eAdv Sci (Weinh) \u003c/em\u003e2023, \u003cstrong\u003e10\u003c/strong\u003e(10):e2204951.\u003c/li\u003e\n\u003cli\u003eLi C, Wu H, Guo L, Liu D, Yang S, Li S, Hua K: \u003cstrong\u003eSingle-cell transcriptomics reveals cellular heterogeneity and molecular stratification of cervical cancer\u003c/strong\u003e. \u003cem\u003eCommun Biol \u003c/em\u003e2022, \u003cstrong\u003e5\u003c/strong\u003e(1):1208.\u003c/li\u003e\n\u003cli\u003eLin S, Sun Y, Cao C, Zhu Z, Xu Y, Liu B, Hu B, Peng T, Zhi W, Xu M\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSingle-nucleus RNA sequencing reveals heterogenous microenvironments and specific drug response between cervical squamous cell carcinoma and adenocarcinoma\u003c/strong\u003e. \u003cem\u003eEBioMedicine \u003c/em\u003e2023, \u003cstrong\u003e97\u003c/strong\u003e:104846.\u003c/li\u003e\n\u003cli\u003ePos O, Radvanszky J, Buglyo G, Pos Z, Rusnakova D, Nagy B, Szemes T: \u003cstrong\u003eDNA copy number variation: Main characteristics, evolutionary significance, and pathological aspects\u003c/strong\u003e. \u003cem\u003eBiomed J \u003c/em\u003e2021, \u003cstrong\u003e44\u003c/strong\u003e(5):548-559.\u003c/li\u003e\n\u003cli\u003eFruman DA, Chiu H, Hopkins BD, Bagrodia S, Cantley LC, Abraham RT: \u003cstrong\u003eThe PI3K Pathway in Human Disease\u003c/strong\u003e. \u003cem\u003eCell \u003c/em\u003e2017, \u003cstrong\u003e170\u003c/strong\u003e(4):605-635.\u003c/li\u003e\n\u003cli\u003eCarragher NO, Frame MC: \u003cstrong\u003eFocal adhesion and actin dynamics: a place where kinases and proteases meet to promote invasion\u003c/strong\u003e. \u003cem\u003eTrends Cell Biol \u003c/em\u003e2004, \u003cstrong\u003e14\u003c/strong\u003e(5):241-249.\u003c/li\u003e\n\u003cli\u003eMetz CE: \u003cstrong\u003eBasic principles of ROC analysis\u003c/strong\u003e. \u003cem\u003eSemin Nucl Med \u003c/em\u003e1978, \u003cstrong\u003e8\u003c/strong\u003e(4):283-298.\u003c/li\u003e\n\u003cli\u003eQuail DF, Joyce JA: \u003cstrong\u003eMicroenvironmental regulation of tumor progression and metastasis\u003c/strong\u003e. \u003cem\u003eNat Med \u003c/em\u003e2013, \u003cstrong\u003e19\u003c/strong\u003e(11):1423-1437.\u003c/li\u003e\n\u003cli\u003ede Visser KE, Joyce JA: \u003cstrong\u003eThe evolving tumor microenvironment: From cancer initiation to metastatic outgrowth\u003c/strong\u003e. \u003cem\u003eCancer Cell \u003c/em\u003e2023, \u003cstrong\u003e41\u003c/strong\u003e(3):374-403.\u003c/li\u003e\n\u003cli\u003eGoenka A, Khan F, Verma B, Sinha P, Dmello CC, Jogalekar MP, Gangadaran P, Ahn BC: \u003cstrong\u003eTumor microenvironment signaling and therapeutics in cancer progression\u003c/strong\u003e. \u003cem\u003eCancer Commun (Lond) \u003c/em\u003e2023, \u003cstrong\u003e43\u003c/strong\u003e(5):525-561.\u003c/li\u003e\n\u003cli\u003eWu Q, You L, Nepovimova E, Heger Z, Wu W, Kuca K, Adam V: \u003cstrong\u003eHypoxia-inducible factors: master regulators of hypoxic tumor immune escape\u003c/strong\u003e. \u003cem\u003eJ Hematol Oncol \u003c/em\u003e2022, \u003cstrong\u003e15\u003c/strong\u003e(1):77.\u003c/li\u003e\n\u003cli\u003eTerry S, Buart S, Chouaib S: \u003cstrong\u003eHypoxic Stress-Induced Tumor and Immune Plasticity, Suppression, and Impact on Tumor Heterogeneity\u003c/strong\u003e. \u003cem\u003eFront Immunol \u003c/em\u003e2017, \u003cstrong\u003e8\u003c/strong\u003e:1625.\u003c/li\u003e\n\u003cli\u003eMortezaee K, Majidpoor J: \u003cstrong\u003eThe impact of hypoxia on immune state in cancer\u003c/strong\u003e. \u003cem\u003eLife Sci \u003c/em\u003e2021, \u003cstrong\u003e286\u003c/strong\u003e:120057.\u003c/li\u003e\n\u003cli\u003eVito A, El-Sayes N, Mossman K: \u003cstrong\u003eHypoxia-Driven Immune Escape in the Tumor Microenvironment\u003c/strong\u003e. \u003cem\u003eCells \u003c/em\u003e2020, \u003cstrong\u003e9\u003c/strong\u003e(4).\u003c/li\u003e\n\u003cli\u003eLiikanen I, Lauhan C, Quon S, Omilusik K, Phan AT, Bartroli LB, Ferry A, Goulding J, Chen J, Scott-Browne JP\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eHypoxia-inducible factor activity promotes antitumor effector function and tissue residency by CD8+ T cells\u003c/strong\u003e. \u003cem\u003eJ Clin Invest \u003c/em\u003e2021, \u003cstrong\u003e131\u003c/strong\u003e(7).\u003c/li\u003e\n\u003cli\u003eFridman WH, Pages F, Sautes-Fridman C, Galon J: \u003cstrong\u003eThe immune contexture in human tumours: impact on clinical outcome\u003c/strong\u003e. \u003cem\u003eNat Rev Cancer \u003c/em\u003e2012, \u003cstrong\u003e12\u003c/strong\u003e(4):298-306.\u003c/li\u003e\n\u003cli\u003eMa S, Zhao Y, Lee WC, Ong LT, Lee PL, Jiang Z, Oguz G, Niu Z, Liu M, Goh JY\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eHypoxia induces HIF1alpha-dependent epigenetic vulnerability in triple negative breast cancer to confer immune effector dysfunction and resistance to anti-PD-1 immunotherapy\u003c/strong\u003e. \u003cem\u003eNat Commun \u003c/em\u003e2022, \u003cstrong\u003e13\u003c/strong\u003e(1):4118.\u003c/li\u003e\n\u003cli\u003eCharoentong P, Finotello F, Angelova M, Mayer C, Efremova M, Rieder D, Hackl H, Trajanoski Z: \u003cstrong\u003ePan-cancer Immunogenomic Analyses Reveal Genotype-Immunophenotype Relationships and Predictors of Response to Checkpoint Blockade\u003c/strong\u003e. \u003cem\u003eCell Rep \u003c/em\u003e2017, \u003cstrong\u003e18\u003c/strong\u003e(1):248-262.\u003c/li\u003e\n\u003cli\u003eJiang P, Gu S, Pan D, Fu J, Sahu A, Hu X, Li Z, Traugh N, Bu X, Li B\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSignatures of T cell dysfunction and exclusion predict cancer immunotherapy response\u003c/strong\u003e. \u003cem\u003eNat Med \u003c/em\u003e2018, \u003cstrong\u003e24\u003c/strong\u003e(10):1550-1558.\u003c/li\u003e\n\u003cli\u003eMortezaee K, Majidpoor J: \u003cstrong\u003eMechanisms of CD8(+) T cell exclusion and dysfunction in cancer resistance to anti-PD-(L)1\u003c/strong\u003e. \u003cem\u003eBiomed Pharmacother \u003c/em\u003e2023, \u003cstrong\u003e163\u003c/strong\u003e:114824.\u003c/li\u003e\n\u003cli\u003eChan TA, Yarchoan M, Jaffee E, Swanton C, Quezada SA, Stenzinger A, Peters S: \u003cstrong\u003eDevelopment of tumor mutation burden as an immunotherapy biomarker: utility for the oncology clinic\u003c/strong\u003e. \u003cem\u003eAnn Oncol \u003c/em\u003e2019, \u003cstrong\u003e30\u003c/strong\u003e(1):44-56.\u003c/li\u003e\n\u003cli\u003eJardim DL, Goodman A, de Melo Gagliato D, Kurzrock R: \u003cstrong\u003eThe Challenges of Tumor Mutational Burden as an Immunotherapy Biomarker\u003c/strong\u003e. \u003cem\u003eCancer Cell \u003c/em\u003e2021, \u003cstrong\u003e39\u003c/strong\u003e(2):154-173.\u003c/li\u003e\n\u003cli\u003eTeicher BA: \u003cstrong\u003eHypoxia and drug resistance\u003c/strong\u003e. \u003cem\u003eCancer Metastasis Rev \u003c/em\u003e1994, \u003cstrong\u003e13\u003c/strong\u003e(2):139-168.\u003c/li\u003e\n\u003cli\u003eKopecka J, Salaroglio IC, Perez-Ruiz E, Sarmento-Ribeiro AB, Saponara S, De Las Rivas J, Riganti C: \u003cstrong\u003eHypoxia as a driver of resistance to immunotherapy\u003c/strong\u003e. \u003cem\u003eDrug Resist Updat \u003c/em\u003e2021, \u003cstrong\u003e59\u003c/strong\u003e:100787.\u003c/li\u003e\n\u003cli\u003eJing X, Yang F, Shao C, Wei K, Xie M, Shen H, Shu Y: \u003cstrong\u003eRole of hypoxia in cancer therapy by regulating the tumor microenvironment\u003c/strong\u003e. \u003cem\u003eMol Cancer \u003c/em\u003e2019, \u003cstrong\u003e18\u003c/strong\u003e(1):157.\u003c/li\u003e\n\u003cli\u003eYang W, Soares J, Greninger P, Edelman EJ, Lightfoot H, Forbes S, Bindal N, Beare D, Smith JA, Thompson IR\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eGenomics of Drug Sensitivity in Cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells\u003c/strong\u003e. \u003cem\u003eNucleic Acids Res \u003c/em\u003e2013, \u003cstrong\u003e41\u003c/strong\u003e(Database issue):D955-961.\u003c/li\u003e\n\u003cli\u003eRavi R, Mookerjee B, Bhujwalla ZM, Sutter CH, Artemov D, Zeng Q, Dillehay LE, Madan A, Semenza GL, Bedi A: \u003cstrong\u003eRegulation of tumor angiogenesis by p53-induced degradation of hypoxia-inducible factor 1alpha\u003c/strong\u003e. \u003cem\u003eGenes Dev \u003c/em\u003e2000, \u003cstrong\u003e14\u003c/strong\u003e(1):34-44.\u003c/li\u003e\n\u003cli\u003eChoudhry H, Harris AL: \u003cstrong\u003eAdvances in Hypoxia-Inducible Factor Biology\u003c/strong\u003e. \u003cem\u003eCell Metab \u003c/em\u003e2018, \u003cstrong\u003e27\u003c/strong\u003e(2):281-298.\u003c/li\u003e\n\u003cli\u003eLi X, Li Y, Liu B, Chen L, Lyu F, Zhang P, He Q, Cheng L, Liu C, Song Y\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eP4HA2-mediated HIF-1alpha stabilization promotes erdafitinib-resistance in FGFR3-alteration bladder cancer\u003c/strong\u003e. \u003cem\u003eFASEB J \u003c/em\u003e2023, \u003cstrong\u003e37\u003c/strong\u003e(4):e22840.\u003c/li\u003e\n\u003cli\u003eSalceda S, Caro J: \u003cstrong\u003eHypoxia-inducible factor 1alpha (HIF-1alpha) protein is rapidly degraded by the ubiquitin-proteasome system under normoxic conditions. Its stabilization by hypoxia depends on redox-induced changes\u003c/strong\u003e. \u003cem\u003eJ Biol Chem \u003c/em\u003e1997, \u003cstrong\u003e272\u003c/strong\u003e(36):22642-22647.\u003c/li\u003e\n\u003cli\u003ePfaendler KS, Tewari KS: \u003cstrong\u003eChanging paradigms in the systemic treatment of advanced cervical cancer\u003c/strong\u003e. \u003cem\u003eAm J Obstet Gynecol \u003c/em\u003e2016, \u003cstrong\u003e214\u003c/strong\u003e(1):22-30.\u003c/li\u003e\n\u003cli\u003eTewari KS, Sill MW, Long HJ, 3rd, Penson RT, Huang H, Ramondetta LM, Landrum LM, Oaknin A, Reid TJ, Leitao MM\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eImproved survival with bevacizumab in advanced cervical cancer\u003c/strong\u003e. \u003cem\u003eN Engl J Med \u003c/em\u003e2014, \u003cstrong\u003e370\u003c/strong\u003e(8):734-743.\u003c/li\u003e\n\u003cli\u003eNordstrom B, Strang P, Lindgren A, Bergstrom R, Tribukait B: \u003cstrong\u003eCarcinoma of the endometrium: do the nuclear grade and DNA ploidy provide more prognostic information than do the FIGO and WHO classifications?\u003c/strong\u003e \u003cem\u003eInt J Gynecol Pathol \u003c/em\u003e1996, \u003cstrong\u003e15\u003c/strong\u003e(3):191-201.\u003c/li\u003e\n\u003cli\u003eAi Y, Zhu H, Xie C, Jin X: \u003cstrong\u003eRadiomics in cervical cancer: Current applications and future potential\u003c/strong\u003e. \u003cem\u003eCrit Rev Oncol Hematol \u003c/em\u003e2020, \u003cstrong\u003e152\u003c/strong\u003e:102985.\u003c/li\u003e\n\u003cli\u003eMcGrail DJ, Pilie PG, Rashid NU, Voorwerk L, Slagter M, Kok M, Jonasch E, Khasraw M, Heimberger AB, Lim B\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eHigh tumor mutation burden fails to predict immune checkpoint blockade response across all cancer types\u003c/strong\u003e. \u003cem\u003eAnn Oncol \u003c/em\u003e2021, \u003cstrong\u003e32\u003c/strong\u003e(5):661-672.\u003c/li\u003e\n\u003cli\u003eTsao MS, Le Teuff G, Shepherd FA, Landais C, Hainaut P, Filipits M, Pirker R, Le Chevalier T, Graziano S, Kratze R\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003ePD-L1 protein expression assessed by immunohistochemistry is neither prognostic nor predictive of benefit from adjuvant chemotherapy in resected non-small cell lung cancer\u003c/strong\u003e. \u003cem\u003eAnn Oncol \u003c/em\u003e2017, \u003cstrong\u003e28\u003c/strong\u003e(4):882-889.\u003c/li\u003e\n\u003cli\u003eGraham K, Unger E: \u003cstrong\u003eOvercoming tumor hypoxia as a barrier to radiotherapy, chemotherapy and immunotherapy in cancer treatment\u003c/strong\u003e. \u003cem\u003eInt J Nanomedicine \u003c/em\u003e2018, \u003cstrong\u003e13\u003c/strong\u003e:6049-6058.\u003c/li\u003e\n\u003cli\u003eSun X, Lv X, Yan Y, Zhao Y, Ma R, He M, Wei M: \u003cstrong\u003eHypoxia-mediated cancer stem cell resistance and targeted therapy\u003c/strong\u003e. \u003cem\u003eBiomed Pharmacother \u003c/em\u003e2020, \u003cstrong\u003e130\u003c/strong\u003e:110623.\u003c/li\u003e\n\u003cli\u003eTao J, Yang G, Zhou W, Qiu J, Chen G, Luo W, Zhao F, You L, Zheng L, Zhang T\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eTargeting hypoxic tumor microenvironment in pancreatic cancer\u003c/strong\u003e. \u003cem\u003eJ Hematol Oncol \u003c/em\u003e2021, \u003cstrong\u003e14\u003c/strong\u003e(1):14.\u003c/li\u003e\n\u003cli\u003eSerganova I, Doubrovin M, Vider J, Ponomarev V, Soghomonyan S, Beresten T, Ageyeva L, Serganov A, Cai S, Balatoni J\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eMolecular imaging of temporal dynamics and spatial heterogeneity of hypoxia-inducible factor-1 signal transduction activity in tumors in living mice\u003c/strong\u003e. \u003cem\u003eCancer Res \u003c/em\u003e2004, \u003cstrong\u003e64\u003c/strong\u003e(17):6101-6108.\u003c/li\u003e\n\u003cli\u003eLjungkvist AS, Bussink J, Kaanders JH, Rijken PF, Begg AC, Raleigh JA, van der Kogel AJ: \u003cstrong\u003eHypoxic cell turnover in different solid tumor lines\u003c/strong\u003e. \u003cem\u003eInt J Radiat Oncol Biol Phys \u003c/em\u003e2005, \u003cstrong\u003e62\u003c/strong\u003e(4):1157-1168.\u003c/li\u003e\n\u003cli\u003eLjungkvist AS, Bussink J, Kaanders JH, van der Kogel AJ: \u003cstrong\u003eDynamics of tumor hypoxia measured with bioreductive hypoxic cell markers\u003c/strong\u003e. \u003cem\u003eRadiat Res \u003c/em\u003e2007, \u003cstrong\u003e167\u003c/strong\u003e(2):127-145.\u003c/li\u003e\n\u003cli\u003eCui YL, Wang X, Li XF: \u003cstrong\u003e(18)F-fluoromisonidazole PET reveals spatial and temporal heterogeneity of hypoxia in mouse models of human non-small-cell lung cancer\u003c/strong\u003e. \u003cem\u003eFuture Oncol \u003c/em\u003e2015, \u003cstrong\u003e11\u003c/strong\u003e(20):2841-2849.\u003c/li\u003e\n\u003cli\u003eLiu Y, Wu J, Huang W, Weng S, Wang B, Chen Y, Wang H: \u003cstrong\u003eDevelopment and validation of a hypoxia-immune-based microenvironment gene signature for risk stratification in gastric cancer\u003c/strong\u003e. \u003cem\u003eJ Transl Med \u003c/em\u003e2020, \u003cstrong\u003e18\u003c/strong\u003e(1):201.\u003c/li\u003e\n\u003cli\u003eGao F, Wang ZZ, Gu JJ, Zhang XJ, Wang HX: \u003cstrong\u003eA Hypoxia-Associated Prognostic Gene Signature Risk Model and Prognosis Predictors in Gliomas\u003c/strong\u003e. \u003cem\u003eFrontiers in Oncology \u003c/em\u003e2021, \u003cstrong\u003e11\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eXie SQ, Ding B, Wang SY, Zhang X, Yan WJ, Xia QQ, Meng D, Shen SY, Yu BJ, Liu HH\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eConstruction of a hypoxia-immune-related prognostic model and targeted therapeutic strategies for cervical cancer\u003c/strong\u003e. \u003cem\u003eInt Immunol \u003c/em\u003e2022, \u003cstrong\u003e34\u003c/strong\u003e(7):379-394.\u003c/li\u003e\n\u003cli\u003eLi QX, Jin LM, Jin M: \u003cstrong\u003eNovel Hypoxia-Related Gene Signature for Risk Stratification and Prognosis in Hepatocellular Carcinoma\u003c/strong\u003e. \u003cem\u003eFront Genet \u003c/em\u003e2021, \u003cstrong\u003e12\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eDi Giovannantonio M, Hartley F, Elshenawy B, Barberis A, Hudson D, Shafique HS, Allott VES, Harris DA, Lord SR, Haider S\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eDefining hypoxia in cancer: A landmark evaluation of hypoxia gene expression signatures\u003c/strong\u003e. \u003cem\u003eCell Genom \u003c/em\u003e2025, \u003cstrong\u003e5\u003c/strong\u003e(2):100764.\u003c/li\u003e\n\u003cli\u003eHarris BH, Barberis A, West CM, Buffa FM: \u003cstrong\u003eGene Expression Signatures as Biomarkers of Tumour Hypoxia\u003c/strong\u003e. \u003cem\u003eClin Oncol (R Coll Radiol) \u003c/em\u003e2015, \u003cstrong\u003e27\u003c/strong\u003e(10):547-560.\u003c/li\u003e\n\u003cli\u003eRajapakse M, Tan J, Rajapakse J: \u003cstrong\u003eColor channel encoding with NMF for face recognition\u003c/strong\u003e. \u003cem\u003eIeee Image Proc \u003c/em\u003e2004:2007-2010.\u003c/li\u003e\n\u003cli\u003eKaluz S, Zhang Q, Kuranaga Y, Yang H, Osuka S, Bhattacharya D, Devi NS, Mun J, Wang W, Zhang R\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eTargeting HIF-activated collagen prolyl 4-hydroxylase expression disrupts collagen deposition and blocks primary and metastatic uveal melanoma growth\u003c/strong\u003e. \u003cem\u003eOncogene \u003c/em\u003e2021, \u003cstrong\u003e40\u003c/strong\u003e(33):5182-5191.\u003c/li\u003e\n\u003cli\u003eZhu M, Peng R, Liang X, Lan Z, Tang M, Hou P, Song JH, Mak CSL, Park J, Zheng SE\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eP4HA2-induced prolyl hydroxylation suppresses YAP1-mediated prostate cancer cell migration, invasion, and metastasis\u003c/strong\u003e. \u003cem\u003eOncogene \u003c/em\u003e2021, \u003cstrong\u003e40\u003c/strong\u003e(41):6049-6056.\u003c/li\u003e\n\u003cli\u003eLin J, Jiang L, Wang X, Wei W, Song C, Cui Y, Wu X, Qiu G: \u003cstrong\u003eP4HA2 Promotes Epithelial-to-Mesenchymal Transition and Glioma Malignancy through the Collagen-Dependent PI3K/AKT Pathway\u003c/strong\u003e. \u003cem\u003eJ Oncol \u003c/em\u003e2021, \u003cstrong\u003e2021\u003c/strong\u003e:1406853.\u003c/li\u003e\n\u003cli\u003eChi Z, Wang Q, Wang X, Li D, Tong L, Shi Y, Yang F, Guo Q, Zheng J, Chen Z: \u003cstrong\u003eP4HA2 promotes proliferation, invasion, and metastasis through regulation of the PI3K/AKT signaling pathway in oral squamous cell carcinoma\u003c/strong\u003e. \u003cem\u003eSci Rep \u003c/em\u003e2024, \u003cstrong\u003e14\u003c/strong\u003e(1):15023.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"cervical squamous cell carcinoma, hypoxia, prognosis, scRNA-seq, Bulk RNA-seq, HIF-1α, machine learning","lastPublishedDoi":"10.21203/rs.3.rs-6832842/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6832842/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eCervical cancer, especially cervical squamous cell carcinoma (CSCC), is a major public health issue in low - and middle - income countries, with advanced recurrence and metastasis linked to poor prognosis. It shows significant intra - tumoral phenotypic heterogeneity and plasticity.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe analyzed the single-cell RNA sequencing data of cervical squamous cell carcinoma available in the Comprehensive Gene Expression Database (GEO) and identified the tumor cell subtype exhibiting hypoxic characteristics. We extracted differentially expressed genes (HRDEGs) between this hypoxia-related cluster and other tumor cells. Based on the CSCC bulk RNA sequencing data published in the Cancer Genome Atlas (TCGA), this subtype was identified to be closely associated with poor prognosis in CSCC.101 combinations consisting of 10 machine learning were used for screening prognostic biomarkers in HRDEGs, and a hypoxia signature was established by multivariate COX regression.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eThe hypoxia signature was validated using the GEO external database. Correlation analysis identified the hypoxia signature as significantly associated with hypoxia and tumor invasion, and verified that higher hypoxia signature are closely related to poorer immune infiltration and responses of immunotherapy and chemotherapy. In addition, the key gene P4HA2 in the hypoxia signature has been demonstrated to be associated with the malignant phenotypes of tumor cells and the regulation of HIF-1α stability.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eOverall, this hypoxia signature is a promising independent prognostic factor, provides new biomarkers for the prognosis of CSCC and a good reference for personalized and precision medicine.\u003c/p\u003e","manuscriptTitle":"A P4HA2-dominated hypoxia signature enables response stratification of multi-therapeutic response in cervical cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-17 11:40:08","doi":"10.21203/rs.3.rs-6832842/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-10-16T06:27:15+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-10-13T19:05:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"303004224326708772162322300275476093246","date":"2025-09-22T10:46:50+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"117143895759428438060582824632178112961","date":"2025-07-22T09:39:44+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-15T09:09:44+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"225895731688567809472964057040009117081","date":"2025-07-15T07:53:32+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-06-15T15:01:48+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-06-15T14:56:14+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-06-10T13:08:21+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-06-10T12:53:16+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Cancer","date":"2025-06-10T12:49:51+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-cancer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bcan","sideBox":"Learn more about [BMC Cancer](http://bmccancer.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bcan/default.aspx","title":"BMC Cancer","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5899d69b-ea8f-4c1f-8df6-d2b8bf1b2004","owner":[],"postedDate":"June 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2026-01-26T16:04:37+00:00","versionOfRecord":{"articleIdentity":"rs-6832842","link":"https://doi.org/10.1186/s12885-026-15597-z","journal":{"identity":"bmc-cancer","isVorOnly":false,"title":"BMC Cancer"},"publishedOn":"2026-01-21 15:57:31","publishedOnDateReadable":"January 21st, 2026"},"versionCreatedAt":"2025-06-17 11:40:08","video":"","vorDoi":"10.1186/s12885-026-15597-z","vorDoiUrl":"https://doi.org/10.1186/s12885-026-15597-z","workflowStages":[]},"version":"v1","identity":"rs-6832842","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6832842","identity":"rs-6832842","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}