Identication and Validation of a Hypoxia-related lncRNA Signature for Prognostic Prediction in Endometrial Cancer

Background: Endometrial cancer (EC) is one of the most common types of gynecological cancer. Hypoxia is an important clinical feature and regulates various tumor processes. However, the prognostic value of hypoxia-related lncRNA in EC remains to be further elucidated. Here, we aimed to characterize the molecular features of EC by the development of a classication system based on the expression prole of hypoxia-related lncRNA. Methods: Univariate Cox regression analysis was used to identify hypoxia-related lncRNAs associated with overall survival. The least absolute shrinkage and selection operator (LASSO) Cox regression model was utilized to construct gene signature. Multivariate Cox regression analysis and receiver operating characteristic (ROC) curve analysis were also performed. Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEEG) pathway, and Gene Set Enrichment Analysis (GESA) were used to identify hypoxia-related lncRNA pathway. Western blot and real-time PCR were used to detect target gene expression. The cell proliferation was determined by using WST-1 assay. Results: Based on univariate Cox regression analysis, we identied 17 hypoxia-related lncRNAs signicantly associated with overall survival. Next, the least absolute shrinkage and selection operator (LASSO) Cox regression model was utilized to construct a multigene signature in the TCGA EC cohort. The risk score was conrmed as an independent predictor for overall survival in multivariate Cox regression analysis and receiver operating characteristic (ROC) curve analysis. Besides, the survival time of EC patients in different risk group was signicantly correlated to clinicopathologic factors, such as age, stage and grade. Furthermore, hypoxia-related lncRNA associated with the high-risk group were involved in various aspects of the malignant progression of EC via Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEEG) pathway, and Gene Set Enrichment Analysis (GESA). Besides, using CIBERSORT analysis, we found a different immune cell environment characterization of EC between different cluster and risk group. Moreover, the risk score was closely correlated to immunotherapy response, Relationships between the risk score and factors. lncRNA

Endometrial cancer (EC) is one of the most common gynecological cancer in developed countries [1,2]. EC has been considered as a global threat to women's health and well-being. Most EC patients are diagnosed at an early stage with a favorable prognosis. However, 20-30% of EC patients diagnosed with a high-risk of recurrence and poor prognosis. EC patients often showed different prognoses and treatment responses, even if with the same degree of progression [3][4][5]. Therefore, effective EC biomarkers must be found to assess prognosis and identify EC patients with high-risk.
Hypoxia is a common phenomenon in solid tumors. Hypoxia is harmful to cancer cells, but it can promote cancer cells to adapt, thus promoting malignant progression [6]. In the hypoxic microenvironment, tumor cells exhibit activating downstream genes to facilitate tumor growth, with strong aggressiveness and metastasis ability [7,8]. In response to hypoxia, the major component of hypoxia signaling pathways is hypoxia inducible factors-1 (HIF-1), which is a heterodimer composed of α and β subunits [9,10]. HIF-1α mainly interact with HIF-1β subunit to form a stable HIF-1 complex in the nucleus. Under hypoxic conditions, the HIF-1 complex binds to the target gene's promoter to induce their transcription. Most of the HIF-1 complex dependent genes were associated with proliferation, epithelial to mesenchymal transition, angiogenesis, and metastasis [11][12][13]. However, due to the lack of effective biomarkers, how hypoxia leads to tumor progression remain to be further elucidated.
Long non coding RNAs (lncRNAs) are a large class of heterogeneous transcripts with a length of more than 200 nucleotides, and their protein coding potential is limited. LncRNAs are widely transcribed in eukaryotic genomes, suggesting their important regulatory role in complex organisms [14,15]. Although only a small number of functional lncRNAs have been well characterized, more and more evidence showed that lncRNAs play a key role in controlling a large number of cancer-related cell processes, such as proliferation, migration, invasion, autophagy and stemness [16][17][18][19]. A speci c group of lncRNAs is regulated by tumor microenvironment, such as hypoxia. The hypoxia responsive lncRNAs, such as NORAD, RAB11B-AS1 and LncHIFCAR, may be the basis of cancer cell survival and disease progression [20][21][22]. MALAT1 was the rst reported lncRNA mediating hypoxia-induced pro-survival autophagy in endometriosis [23]. However, there is few reports about the relationship between hypoxia related lncRNA and EC in recent years. Whether lncRNAs are involved in the response to hypoxia in EC, and the prognostic value of hypoxia-related lncRNA in EC remains to be further elucidated.
In this study, we rst downloaded the RNA expression pro les of hypoxia-related lncRNA and corresponding clinical data of EC patients from public databases. The different lncRNA expression patterns were detected among EC cases, to identify the candidate lncRNA biomarkers based on The Cancer Genome Atlas data (TCGA). Besides, we found that the hypoxia-related lncRNA could classify the EC patients with signi cantly different clinical and molecular characteristics. Finally, we focused on lncRNA SOS1-IT1, and found it was signi cantly upregulated in EC cells under hypoxic conditions. We evaluated its biological role and clinical signi cance in EC progression and revealed SOS1-IT1 is hypoxiainducible and directly transactivated by HIF-1α.

Data collection
The TCGA RNA sequencing dataset and corresponding clinicopathological characteristics information, such as age, grade, stage, radiation therapy, surgical approach and survival information were downloaded from TCGA database (http://www.cancergenome.nih.gov/).
Identi cation of differentially expressed hypoxia-related lncRNA Based on the Spearman correlation analysis, lncRNAs related to the hypoxia were identi ed (Filter: |r| > 0.5 and P < 0.001). The prognosis-related lncRNAs were screened by K-M survival analyses (P < 0.01). Differentially expressed hypoxia-related lncRNA were identi ed in EC from the TCGA datasets by using the EdgeR package in R statistical software. The signi cance criteria for determining differentially expressed lncRNA were set as adjusted P value < 0.05. Heatmaps were plotted by the "pheatmap" package.

Comprehensive analysis of interaction network
According to the correlation e ciency and probability cut-off value (Filter: |r| > 0.5 and P < 0.001), 677 lncRNAs were screened out as hypoxia-related lncRNAs. Then we explored the interaction network between hypoxia and hypoxia-related lncRNAs, by using the Cytoscape software.
Generation of hypoxia-related lncRNA signature The expression data of hypoxia-related lncRNA in 550 EC and 35 normal tissues were analyzed with the Limma package and visualized as a heatmap. We carried out consensus clustering with the R package "ConsensusClusterPlus." Then we used "Ggplot2" and "Limma" package for PCA analysis. The "Glmnet" and "Survival" packages were used to perform LASSO regression analysis [24]. We also performed the univariate and multivariate Cox hazard analysis of clinicopathological characteristics information by "survival" package.
Construction of hypoxia-related lncRNA prognostic model Prognosis-related lncRNA were constructed using multivariate cox regression. The risk score of each patient was calculated according to the predictive signature model. Using the median risk score as the cutoff point, EC patients were divided into the high-risk group and low-risk group. The prediction e ciency of the risk signature model was evaluated by the receiver operating characteristic (ROC) curve using R package survival ROC.
Functional enrichment analysis of the hypoxia-related lncRNA.
Functional enrichment analysis of the hypoxia-related lncRNA was conducted using DAVID, including biological functions, cellular components, and molecular functions. The Kyoto Encyclopedia of Genes and Genomes (KEGG) database was searched for signi cant pathways. We also performed Gene set enrichment analysis (GSEA) to identify the signi cantly alerted KEGG pathways between the high-risk and low-risk groups. The Java GSEA program were used and the gene set from Molecular Signatures Database was selected as the reference gene set. Biological processes with normalized p value < 0.05 and false discovery rate (FDR) q value < 0.05 were considered statistically signi cant.

Evaluation of tumor-in ltrating immune cells and the immune in ltration level
The subpopulation of 22 immune cells in each tumor sample was explored by CIBERSORT algorithm in different clusters. The CIBERSORT results of samples with p < 0.05 showed that the inferred fractions of immune cell populations were accurate and could be further analyzed. Furthermore, we analyzed the correlation between tumor immune cell in ltrating and prognostic risk signature.

Tumor mutation burden assessment
The tumor mutation burden (TMB) for each organization was detected by the VarScan method, as calculated using the R package "maftools". According to the cutoff value of median TMB data, EC Patients were divided into high TMB group and low TMB group. We also analyzed the correlation between TMB and prognostic risk signature.

Cell culture
Human endometrial cancer cell lines Ishikawa was purchased from ATCC (American Type Culture Collection). The cells were grown in DMEM supplemented with 10% fetal bovine serum (FBS) and cultured at 37 o C with 5% CO 2 .
Total RNA extraction and Real-time PCR The total RNA was extracted by TRIZOL reagent (Invitrogen) according to the manufacturer's instruction. The cDNA was obtained from the puri ed RNA using a PrimeScript RT Reagent Kit (Takara). Real-time PCR was performed using the SYBR Premix Ex Taq (Takara) following the manufacturer's instructions. The results were normalized to β-actin gene.

Cell proliferation assays
The cell proliferation assays were determined using a commercial WST-1 assay kit according to the manufacturer's instructions as described before [25][26][27]. First, 15 μl WST-1 reagent was added to the cells in the 96-well plates, and incubated for 2 h. Later, the OD value was measured in a microplate reader at 450 nm.

Dual-luciferase reporter assays
For the luciferase reporter experiment, the SOS1-IT1 wild type (WT) or mutant vectors were used. The Ishikawa cells were seeded in six-well plates and transfected with relative plasmids or siRNA for 48 h. The Dual-Luciferase Reporter Assay System was used to assess the activities of re y luciferase and Renilla luciferase sequentially. Then the relative luciferase activities were calculated, and control cells were used for normalization.

Statistical analysis
Differences between survival curves were generated by the Kaplan-Meier method and compared by logrank tests. The multivariate analysis was performed using the cox proportional hazard model. For comparisons of two groups, a t-test was used. R studio package was used for all statistical analysis. All statistical tests were only considered statistically signi cant when p <0.05 was achieved.

Results
Identi cation hypoxia-related lncRNA in EC To better understand the important role of hypoxia-related lncRNA in oncogenesis and progression, we rst downloaded from the hallmark gene sets of hypoxia including 200 genes from Molecular Signatures Database. RNA-seq data from 550 tumor tissue samples and 35 normal samples were downloaded from TCGA. According to the correlation e ciency and probability cut-off value (Filter: |r| > 0.5 and P < 0.001), 677 lncRNAs were screened out as hypoxia-related lncRNA. In order to investigate the prognostic value of these hypoxia-related lncRNA in EC, univariate Cox regression analysis was performed based on the expression levels of these lncRNA in TCGA database. As a result, we found that 17 out of the 677 lncRNAs were signi cantly associated with overall survival (p < 0.05) ( Figure 1A). The expression values of 17 hypoxia-related lncRNAs were extracted from EC patients. The expression pro le of 17 prognostic associated hypoxia-related lncRNA was showed in the heatmap and box plot (Figure 1B-C). As shown in Figure 1C, 17 hypoxia-related lncRNAs were signi cantly abnormally expressed in EC tissues samples.

Classi cation of EC based on hypoxia-related lncRNA
In order to analysis the consensus cluster of hypoxia-related lncRNA in EC, we used the common clusterplus package to identify the different groups of hypoxia-related lncRNA based on their coexpression patterns in EC tissues from the TCGA database. Due to the grouping was suboptimal when they were divided into more than 2 clusters, we divided the hypoxia-related lncRNA into two groups based on their expression indices using k = 2 as the optimal value (Figure 2A-D). Next, we analyzed the relationship between these two clusters and the clinicopathological characteristics of EC patients. We found that consensus clustering could make signi cant differences in the clinical and molecular characteristics of the two EC clusters ( Figure 2E). Cluster 1 patients were strikingly correlated with stage, grade, fustat and disease type by Chi-square test. Besides, compared with patients in cluster 2, EC patients in cluster 1 showed a shorter overall survival time ( Figure 2F).
Prognostic value of hypoxia-related lncRNA and construction of a risk signature predicting prognosis To identi ed the most powerful prognostic hypoxia-related lncRNAs, the last absolute shrinkage and selection operator (LASSO) Cox regression analysis to the 17 prognosis-related lncRNAs was conducted. We constructed prognostic models using the multivariate Cox proportional hazards regression analysis and the coe cient of each independent prognostic gene were calculated. The risk score was estimated based on the coe cients from the LASSO results. According to the median risk score, EC patients were assigned into low-risk and high-risk groups. The distribution of the hypoxia-related risk signature in the TCGA dataset and survival status of EC patients in different groups were shown in Figure  3A-B. After adjusting for clinicopathological features such as age, grade, stage, radiation therapy, surgical approach, and disease type, we found that age, grade, stage, disease type and risk score were correlated with the OS of EC patients in univariate analysis, while multivariate COX regression analysis showed that age, grade, stage, radiation therapy, surgical approach and risk score were independent risk factors for the prognosis of EC patients ( Figure 3C-D).
The heatmap of these most signi cantly six lncRNAs were shown in Figure 3E, and we observed a strong correlation between the risk score and the clinicopathological characteristics such as stage, age and grade. The result of survival analysis showed that the high-risk group had signi cantly shorter survival time compared to low-risk group ( Figure 3F). The AUC value of the ROC curve in risk score for 5-year survival is 0.691, which is obviously higher than that of ROC in grade (0.662), radiation therapy (0.371), surgical approach (0.461) and disease type (0.569), but lower than stage (0.702) ( Figure 3G). In addition, the calibration plot revealed ideal agreements of these genes between the actual observations and 1-, 3and 5-year OS predicted by the nomogram ( Figure 3H). These results indicated that the prognostic index based on hypoxia-related lncRNA has the potential to predict the survival of EC patients.
Relationships between the risk score and clinicopathologic factors Next, we built a complete prognostic model based on the entire set. The strati cation analysis was done according to the age, grade, stage, radiation therapy and surgical approach. EC patients were strati ed into age ≤60 and > 60 subgroup, grade G1-G2 and G3-G4 subgroup, stage I-II and III-IV subgroup, surgical approach minimally invasive and open subgroup, radiation therapy No and Yes subgroup. For the EC patients in age > 60 subgroup, grade G3-G4 subgroup, and stage III-IV subgroup, the survival time of patients was signi cantly shorter than that of patients in another group ( Figure 4A), and the average risk score was much higher (Figure 4D-E). However, there is no difference between surgical approach and radiation therapy subgroup. Besides, we found the survival time of patients in high-risk group was signi cantly shorter in age > 60 subgroup, grade G3-G4 subgroup, stage III-IV subgroup, surgical approach minimally invasive subgroup, and radiation therapy No subgroup ( Figure 4B-C).
Biological characteristics and pathway analysis of hypoxia-related risky lncRNA The EC patients in TCGA were divided into high-risk and low-risk groups based on the median risk score. Then, the lncRNA signi cantly upregulated (fold change> 1 and p < 0.05) or downregulated (fold change < −1 and p < 0.05) were selected for Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEEG) pathway, and Gene Set Enrichment Analysis (GESA). To elucidate the biological functions and pathways that were associated with the risk score, the hypoxia-related lncRNA between the high-risk and low-risk groups were used to perform GO enrichment and KEGG pathway analyses ( Figure 5A-D). As expected, hypoxia-related lncRNAs were enriched in several cancer-related biological pathways, such as membrane potential regulation, cell-cell adhesion, and protein digestion signaling pathway, et al. The GSEA analysis results showed that different risk group was involved in multiple signi cantly enriched pathways, including cell cycle, wnt signaling, spliceosome, et al ( Figure 5E). These results showed the two-risk group identi ed based on the six hypoxia-related risky lncRNAs were closely associated with the malignancy of EC.

The correlation of risk score with immune cell environment characterization in EC
To clarify the potential role of the two clusters (Cluster 1 and 2) in immune cell environment characterization of EC, we investigated the expression levels of the 22 immune cell types in ltration between the two clusters and immune signature by CIBERSORT analysis. As shown in Figure 6A, Cluster 2 patients exhibited higher levels of proportions in CD4 memory resting T cells, activated NK cells and M0 Macrophages. While a higher proportion of CD8 T cells, follicular helper T cells, M1 Macrophages were enriched in Cluster 1 patients. After further uncovering the role of risk score in immune cell environment characterization of EC, we found the risk score was positively correlated with activated dendritic cells, naive B cells, and gamma delta T cell, which meant that hypoxia-related lncRNAs in the signature might in uence the tumor immune microenvironment by promoting activated dendritic cells, naive B cells, and gamma delta T cell to in ltrate into the tumor tissue. However, the neutrophils, resting dendritic cells, regulatory T cell (Tregs), CD8 T cells, and activated NK cells were negatively correlated with risk score ( Figure 6B-I).
The risk score of hypoxia-related lncRNA was associated with immunotherapy, microsatellite instability and tumor mutation burden in EC The combination of anti-CTLA-4 and anti-PD-1 treatment could increase the proportion of activated CD8+ cells and natural killer cells in the tumor microenvironment, and decrease the proportion of inhibitory immune cells, resulting in changes in the immune landscape, which achieved therapeutic effect in the mouse model and prolonged the tumor free survival [28]. After analyzing the difference in response to immunotherapy between different risk score group. We found that low-risk group tended to respond effectively to immunotherapy such as anti-PD-1 and anti-CTLA-4 therapy ( Figure 7A-D). High microsatellite instability (MSI-H) is associated with the response to immunotherapy treatment. Interestingly, we found that the hypoxia-related lncRNA risk score of EC patients with MSI-H was lower than that of EC patients with low MSI or microsatellite stability (MSS) (Figure 7E-F). Tumor genome mutation leads to the production of new antigens that is bene cial to immunotherapy. Tumor mutation burden (TMB) is an important biomarker, which can be used to predict the response of a variety of tumors to PD-1/PD-L1 targeted immunotherapy. The box plot showed that there was a difference in mutation frequency between different risk score groups ( Figure 7G). We observed that the risk score was negatively correlated with TMB in correlation analysis (R=-0.23, P=1.2e-07) ( Figure 7H). The EC patients were classi ed into the high and low TMB group. Kaplan-Meier analysis showed High TMB signi cantly correlated with better prognosis ( Figure 7I). Then we evaluated the synergistic effect of risk score on prognosis strati cation. The survival analysis showed that there was signi cant difference in survival rate according to risk score between different TMB subgroups ( Figure 7J). SOS1-IT1 is clinically relevant in EC and promotes EC cell growth SOS1-IT1 was the most correlated prognostic lncRNAs in this model. Therefore, we will further evaluate the role of SOS1-IT1 in EC to verify the hypoxia-related lncRNA model. To investigate the clinical signi cance of SOS1-IT1 in EC, we analyzed its expression and clinical relevance in TCGA EC database. As shown in Figure 8A-B, SOS1-IT1 was overexpressed in tumor tissues. Analysis from clinical investigations suggested that the aberrant level of SOS1-IT1 was closely correlated with clinicopathological parameters of EC, including the age, disease type, fustat and grade ( Figure 8C-F).
Kaplan-Meier analysis revealed that high expression of SOS1-IT1 was signi cantly associated with a poor prognosis in EC patients ( Figure 8G). To directly investigate the biological functions of SOS1-IT1 on EC cells, SOS1-IT1 was knocked down or overexpressed respectively in EC cell line Ishikawa cells. The real-time PCR analysis results indicated that the expression level of SOS1-IT1 was markedly decreased or increased in knocking down or overexpressing cells respectively ( Figure 8H-I). The downregulation of SOS1-IT1 signi cantly inhibited EC cell growth. In contrast, overexpressing SOS1-IT1 yielded the opposite results ( Figure 8J-K). These results indicate that SOS1-IT1 may lead to increased EC aggressiveness.

SOS1-IT1 is upregulated under hypoxia and directly transactivated by HIF-1α
To further con rm whether SOS1-IT1 was a functional effector of hypoxia in EC progression, we treated Ishikawa cells with hypoxia or its chemical inducer CoCl 2 for 24 h, and found the expression level of SOS1-IT1 was signi cantly increased ( Figure 9A). Besides, the expression level of SOS1-IT1 was signi cantly decreased after knocking down HIF-1α, which is the main signaling pathway response component to hypoxia, both in normoxia and hypoxia condition (Figure 9 B-C). After detecting of the upstream region (~1KB upstream) of SOS1-IT1 gene by promoter sequence analysis, we found a putative HIF-1α binding site in the promoter region of SOS1-IT1 gene ( Figure 9D). We generated the mutant binding site of the reporter constructs, following with luciferase assays after transfection of different reporter constructs in EC cells. Under normoxia conditions, there was no much difference of luciferase activity in SOS1-IT1 wild type (WT) or mutant, as well as empty vector group. However, under hypoxic conditions, there was a nearly six-fold induction of luciferase activity in SOS1-IT1 WT construct. Mutation in SOS1-IT1 promoter decreased the luciferase activity to nearly basal levels caused by hypoxia ( Figure  9E). Besides, the Chromatin immunoprecipitation (ChIP) assays con rmed that HIF-1α directly bind to the promoter region promoter of SOS1-IT1 ( Figure 9F). Next, we knocked down HIF-1α, and found HIF-1α suppression remarkably repressed the luciferase density in cells with WT promoter, but not in mutant group ( Figure 9G). Moreover, SOS1-IT1 silencing partially reduced EC cell proliferation ability by hypoxia condition or HIF-1α overexpression ( Figure 9H-J). Totally, these results intensively indicated that SOS1-IT1 was upregulated under hypoxia and a direct transcriptional target of HIF-1α.

Discussion
Endometrial cancer is the sixth most common neoplasm in females, with rapidly increasing in the worldwide and causes ∼74,000 deaths per year [1,3]. The TCGA networking group initially identi ed four molecular subtypes with different prognosis based on genome somatic copy number changes, microsatellite instability and tumor mutation load [29,30]. There are more and more molecular targeted therapies and diagnostic methods for EC [31]. However, the clinical effect of EC is still unsatisfactory. In recent years, many studies have shown that the lncRNA were closely associated with the prognosis of EC patients. In this study, we established a hypoxia-related lncRNA signature, which can effectively distinguish EC patients and predict their survival. To the best of our knowledge, this is the rst study to investigate the association of hypoxia-related lncRNA with prognostic features in EC patients.
The hypoxic microenvironment common to cancer cells emerges as an important factor for cancer progression [6,7]. Besides, hypoxia is an aggressive feature in EC, which can be used for treatments targeting biological changes related to hypoxia [32]. Growing evidence indicates a close correlation between lncRNA and EC [19]. However, there is no systematic study on the correlation between hypoxia and EC. The rapid development of high-throughput gene sequencing technology has laid a foundation for the research of large cancer data [33]. A large number of genomic data were extracted from a single specimen to identify new prognostic and pharmacological biomarkers [34,35]. This selected risk prediction may be a more targeted and effective prognostic assessment for predicting positive clinical outcomes. Compared with other known prognostic indicators, it may be a more effective classi cation tool for EC patients.
In the current study, we built a hypoxia-related lncRNA signature. We found that age, grade, stage, disease type and risk score were correlated with the OS of EC patients, and were independent risk factors for the prognosis of EC patients. In order to understand the mechanism of these signature, we used GSEA to analyze the KEGG pathway between different risk groups. We found that the most signi cant enrichment pathways in the high-risk group were cell cycle and wnt signaling, which were reported associated with the poor prognosis of EC. This partly explained the molecular mechanism of the difference in prognosis between high-risk group and low-risk group. Besides, using CIBERSORT analysis, we found a different immune cell environment characterization of EC between different cluster and risk group. In addition, the risk score was closely related to immunotherapy response, microsatellite instability and TMB.
In order to further con rm the model built in bioinformatics analysis, we select SOS1-IT1 to validate its role in EC cell line. SOS1-IT1 was rst reported as the core lncRNA in the prognostic model of ivermectinrelated three-lncRNA signature in ovarian cancer [36]. It was also identi ed as autophagy-related lncRNA in endometrial cancer [37]. However, its detailed function and molecular mechanism in EC need to be further elucidated. In this study, we found SOS1-IT1 was overexpressed in tumor tissues, and closely correlated with clinicopathological parameters of EC, including the age, disease type, fustat and grade. Moreover, we detected SOS1-IT1 regulated EC cell growth by using knocked down and overexpression experiments. Interestingly, we also found SOS1-IT1 was a functional effector of hypoxia in EC progression. Its expression level was increased in hypoxia or its chemical inducer CoCl 2 . Additionally, the important hypoxia regulatory factor HIF-1α can directly bind SOS1-IT1 promoter region, and affect its expression level. These results further indicate the relationship between SOS1-IT1 and hypoxia, which may increase EC aggressiveness.

Conclusion
In conclusion, we rst clari ed the clinical signi cance of hypoxia-related lncRNA signature in predicting the overall survival rate of EC patients. The hypoxia-related lncRNA was involved in the growth and progression of EC through different pathways. Consistent with previous studies, our ndings suggested that some hypoxia-related lncRNA can predict survival outcomes and monitor tumor progression, which increase the reliability of hypoxia-related lncRNA signature. It is also of great signi cance to reveal the potential molecular mechanism and roles of these lncRNA in other types of malignant tumors. The data and materials can be found from the rst author and corresponding author.

Ethics approval and consent to participate
This study was approved by the Ethics Committee of The Third A liated Hospital of Zhengzhou University.

Consent for publication
All listed authors have actively participated in the study and have read and approved the submitted manuscript.      The risk score of hypoxia-related lncRNA was associated with immunotherapy, microsatellite instability and tumor mutation burden in EC. Kaplan-Meier analysis showed the correlation between tumor mutation burden and EC prognosis (I), and the difference in survival rate according to risk score between different TMB subgroups (J).