Development of a Novel Prognostic Model for Esophageal Squamous Cell Carcinoma: Insights into Immune Cell Interactions and Drug Sensitivity

Development of a Novel Prognostic Model for Esophageal Squamous Cell Carcinoma: Insights into Immune Cell Interactions and Drug Sensitivity | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Development of a Novel Prognostic Model for Esophageal Squamous Cell Carcinoma: Insights into Immune Cell Interactions and Drug Sensitivity Pu Wang, Bin Du, Lingyu Wei, Jia Wang, Jinshang Wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3208103/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 14 Apr, 2024 Read the published version in Cancer Investigation → Version 1 posted You are reading this latest preprint version Abstract Esophageal squamous cell carcinoma (ESCC) is a highly aggressive upper gastrointestinal tumor with a 5-year survival rate of less than 20%. Therefore, developing new effective prognostic markers is of great clinical significance. In this study, we utilized datasets specific to ESCC and analyzed differentially expressed genes in each dataset. By conducting Venn analysis, we identified genes that exhibited significant differential expression across multiple datasets. Through gene interaction network analysis, we identified a core set of genes (23 genes) and established a prognostic model for ESCC using the COX algorithm (p=0.000245, 3-year AUC=0.98). The high-risk group of patients showed a significantly worse prognosis compared to the low-risk group. Furthermore, immune interaction network analysis revealed a strong association between increased risk values and an elevated presence of M2 macrophages within tumor tissues. Drug sensitivity analysis indicated that the high-risk group of patients exhibited poorer sensitivity to first-line chemotherapy drugs for ESCC. Notably, there was a significant positive correlation between the expression of core genes and immune checkpoint genes such as SIGLEC15, PDCD1LG2, and HVCR2. The high-risk group exhibits decreased Tumor Immune Dysfunction and Exclusion (TIDE) values, indicating that immune checkpoint blockade therapy might result in more favorable outcomes for these individuals. The immune checkpoint blockade (ICB) therapy may potentially yield better outcomes for these patients. In summary, through comprehensive bioinformatics analysis, we have established a highly effective prognostic model consisting of 23 genes for ESCC. An increased risk score in this model indicates a stronger infiltration of M2 macrophages and poorer sensitivity to chemotherapy drugs. Moreover, immune checkpoint blockade therapy may hold greater benefits for patients in the high-risk group. Biological sciences/Cancer Biological sciences/Cancer/Cancer genomics Biological sciences/Cancer/Cancer therapy Biological sciences/Cancer/Gastrointestinal cancer Biological sciences/Cancer/Tumour biomarkers Esophageal squamous cell carcinoma gene signature prognosis immunotherapy Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Introduction Esophageal cancer (EC) is ranked as the eighth most common cancer and the sixth leading cause of cancer-associated mortality, with a low five-year overall survival rate[1]. Base on pathological characteristics, EC is typically categorized into Esophageal Squamous Cell Carcinoma (ESCC) and Esophageal Adenocarcinoma (EAC). The characteristics and causes of EC may vary by regions or races. In China, ESCC accounts for 90-95% of all esophageal cancer incidences[2]. Smoking, alcohol consumption, consumption of overheated water, and other environmental factors are recognized as key risk elements for esophageal squamous cell carcinoma[3]. Recent advancements in surgical techniques and the use of neoadjuvant chemotherapy have significantly improved the prognosis for patients with esophageal squamous cell carcinoma. However, as most patients with esophageal cancer are diagnosed in the late stages, the five-year survival rate for patients with esophageal squamous cell carcinoma still remains less than 20%[4]. Therefore, the development of molecular markers that can guide the prognosis and treatment plan for ESCC patients is of great clinical significance. The advancement of bioinformatics in recent years has facilitated the development of gene signatures based on the expression of multiple genes, offering new directions for molecular subtyping and treatment of tumors[5]. However, this field of study has not seen significant progress in esophageal cancer, largely due to insufficiencies in data sets. In existing studies, researchers have utilized multiple datasets and examined genes associated with signal pathways, including ferroptosis [6, 7] , epithelial-mesenchymal transition [8], and autophagy-related genes [9]. Nevertheless, due to various factors such as lack of data sets, inferior quality of data sets, and incompleteness of research schemes, the established gene signatures have not shown substantive discriminative capability in prognosticating for ESCC patients, with AUC values often distributed around 0.7. Consequently, the currently established gene signatures remain less practicable. In our study, we utilized multiple ESCC datasets encompassing paired adjacent non-cancerous and cancerous tissues. We analyzed their common differential genes, and based on this, we established a multi-gene marker composed of core genes, which we subsequently optimized. Finally, through a wide variety of analyses, we unraveled the inherent molecular mechanisms through which these gene markers can efficiently predict the prognosis of ESCC patients. Materials and Methods 1.1 Data acquisition The available datasets related to esophageal squamous cell carcinoma (ESCC) are mainly obtained from the Gene Expression Omnibus (GEO, www.ncbi.nlm.nih.gov/geo/) and The Cancer Genome Atlas (TCGA, www.cancer.gov/) databases. The GSE22954, GSE77861, and GSE20347 datasets are obtained from the GEO database. The ESCC dataset with prognosis information is sourced from the TCGA database. The inclusion criteria for data sets are: (a)having a sufficient number of esophageal squamous cell carcinoma samples; (b) having paired adjacent cancer tissue or healthy tissue samples. 1.2 Acquisition of Differentially Expressed Genes We obtained RNA-sequencing expression (level 3) profiles and associated clinical data for ESCC from the GEO and TCGA datasets. The limma package in R software was used to analyze the differentially expressed mRNA. The threshold for differential expression of mRNAs was defined as “Adjusted P 1 or Log2(Fold Change)< −1.” 1.3 Analysis of shared differentially expressed genes To the identification of shared differentially expressed genes across multiple gene sets, we employed VEN analysis. The construction of Venn diagrams was performed using the JVENN tool [10]. 1.4 Pathway and process enrichment analysis Pathway and process enrichment analysis were performed at Metascape (metascape.org) [11]. The enrichment analysis for each pathway and process was executed using given gene lists and the following ontology sources: KEGG Pathway, GO Biological Processes, Reactome Gene Sets, Canonical Pathways, CORUM, WikiPathways, and PANTHER Pathway. The enrichment background used all genes in the genome. Terms gathered included those with a p-value of below 0.01, a minimum of 3 counts, and an enrichment factor above 1.5, which is the ratio of the observed counts to the expected counts predicted by chance. These terms were then organized into clusters based on their mutual traits. P-values were derived from the cumulative hypergeometric distribution, whilst q-values were determined using the Benjamini-Hochberg procedure to account for any multiple testings[12]. The similarity metric for the hierarchical clustering of the enriched terms applied Kappa scores, considering sub-trees with a similarity above 0.3 as a cluster. The most statistically significant term within a cluster was selected to represent that cluster. 1.5 Pathways network analysis To better understand the links between the terms, we’ve chosen a subset of enriched terms and displayed them in a network plot. Here, terms with a similarity of over 0.3 are linked by edges. From each of the 20 clusters, we’ve specifically chosen the terms with the top p-values. However, we’ve ensured that each cluster doesn’t exceed 15 terms and the total term count doesn’t surpass 250. The network visualization was achieved using Cytoscape. 1.6 Genes relative transcription factor analysis The transcription factor analysis related to differentially expressed genes was performed using TRRUST online analysis [13]. 1.7 Protein-protein interaction enrichment analysis For each given genes, the Molecular Complex Detection (MCODE) algorithm10 were applied to identify densely connected network components[14]. 1.8 Correlation analysis of genes. The genes correlation map is realized by the R software package ggstatsplot, and the multi-gene correlation pheatmap is displayed by the R software package[15]. Spearman’s correlation analysis to describe the correlation between quantitative variables without a normal distribution. P values less than 0.05 were considered statistically significant (*P < 0.05). 1.9 mRNA expression of core genes RNA-sequencing expression (level 3) profiles and corresponding clinical information for ESCC were downloaded from the TCGA dataset(https://portal.gdc.com). All the analysis methods and R package were implemented by R version 4.0.3[16]. 1.10 Establishment of a prognostic model. RNA-sequencing expression (level 3) profiles and corresponding clinical information for xx were downloaded from the TCGA dataset(https://portal.gdc.com). Converting counts data to TPM and normalizing the data log2 (TPM+1), keeping samples with clinical information at the same time. Indicted samples for subsequent analysis. Log-rank test was used to compare differences in survival between these groups. The time-ROC (v 0.4) analysis was used to compare the predictive accuracy of genes and risk score. Multivariate cox regression analysis was used to construct a prognostic model, and the R package survival used for the analysis [17]. 1.11 Analysis of the correlation between gene expression and the amount of immune cells in the tumor. TRNA-sequencing expression (level 3) profiles and corresponding clinical information for ESCC were downloaded from the TCGA dataset (https://portal.gdc.com). The R software GGSTATSPLOT package was used to draw the correlations between gene expression and immune score [18], the R software heatmap package was used to draw multi-gene correlation. Used Spearman’s correlation analysis to describe the correlation between quantitative variables without a normal distribution. P values less than 0.05 were considered statistically significant (*P < 0.05). 1.12 Network analysis of the correlation between the prognostic model and immune cells. To assess the reliable results of immune score evaluation, we used immuneeconv. It’s an R software package that integrates six latest algorithms, including TIMER, xCell, MCP-counter, CIBERSORT, EPIC and quanTIseq. These algorithms had beeen benchmarked, each had a unique advantage. Analysis and visualization are performed by the R software ggClusterNet package (https://doi.org/10.1002/imt2.32)[19]. 1.13 Prediction of drug sensitivity. RNA-sequencing expression (level 3) profiles and corresponding clinical information for ESCC were downloaded from the TCGA dataset (https://portal.gdc.com).Predicted the chemotherapeutic response for each sample based on the largest publicly available pharmacogenomics database the Genomics of Drug Sensitivity in Cancer (GDSC), https://www.cancerrxgene.org/.The prediction process was implemented by R package “pRRophetic”. The samples' half-maximal inhibitory concentration (IC50) was estimated by ridge regression. All parameters were set as the default values using the the batch effect of combat and tissue type of all tissues, and the duplicate gene expression was summarized as mean value[20]. 1.14 Prediction of patients’ sensitivity to immune checkpoint blockade(ICB) therapy. Potential ICB response was predicted with TIDE algorithm[21]. Results 2.1 Identification of signature differentially expressed genes in esophageal squamous cell carcinoma Esophageal cancer has a very poor prognosis, but there is currently no effective prognostic evaluation model in clinical practice. We attempted to identify genes with common differential expression in esophageal squamous cell carcinoma through data analysis. Four datasets with paired cancer adjacent tissues were retrieved from the GEO and TCGA databases, namely GSE22954, GSE77861, GSE20347 (GEO), and the esophageal squamous cell carcinoma dataset in the TCGA database (Fig. 1A). Through VEN analysis, we identified 119 differentially expressed genes with high homogeneity (Fig. 1B). 2.2 Cluster analysis of differentially expressed genes Furthermore, we performed cluster analysis on the 119 genes that showed significant differential expression in different esophageal squamous cell carcinoma databases using the Metascape website. In the pathway cluster analysis, the top five pathways were extracellular matrix organization, burn wound healing, PID integrin1 pathway, regulation of cell cycle process, and network map of SARS-CoV-2 signaling pathway (Fig. 2A). The network analysis of pathway clustering revealed that extracellular matrix organization, burn wound healing, and regulation of cell cycle process pathways predominated among the pathways showing significant clustering (Fig. 2B). Additionally, using the TRRUST module, we analyzed the transcription factors behind these differentially expressed genes. The results showed that the classical transcription factors RELA, NFKB1, and SP1 may play important roles in the expression of these genes (Fig. 2C). The relationship between the transcription factors and the corresponding overlapped genes was listed in Table 1. Table 1. Transcription factors associated with overlapping genes Key TF Description overlapped genes P value Q value List of overlapped genes RELA v-rel reticuloendotheliosis viral oncogene homolog A (avian) 14 4.74E-09 3.41E-07 CXCL8,COL1A2,MMP2,TNFAIP3,CCL20,PLAU,MMP12 , TIMP1,PLAUR,SERPINE1,MMP1,MMP3,FN1,CXCL1 NFKB1 nuclear factor of kappa light polypeptide gene enhancer in B-cells 1 13 4.52E-08 1.63E-06 MMP3,TNFAIP3,MMP1,TIMP1,COL1A2,CXCL8,PLAU , MMP2,FN1,PLAUR,CCL20,SERPINE1,CXCL1 SP1 Sp1 transcription factor 14 1.20E-06 2.55E-05 CDH3,HMGA2,MMP2,FOXM1,KRT14,PLAU,SERPINE1 , TIMP1,CXCL1,HOXB7,ODC1,SPP1,IGFBP3,PLAUR ETS2 v-ets erythroblastosis virus E26 oncogene homolog 2 (avian) 5 1.42E-06 2.55E-05 MMP1,CXCL8,MMP3,CDK1,MMP2 HDAC1 histone deacetylase 1 6 4.89E-06 7.04E-05 SNAI2,IGFBP3,SPP1,CXCL8,BAMBI,COL1A2 FLI1 Friend leukemia virus integration 1 4 1.31E-05 0.000157 MMP1,COL1A2,IGFBP3,FOXM1 E2F1 E2F transcription factor 1 7 1.90E-05 0.000174 SERPINE1,CDK1,FOXM1,CDC6,AURKA,NELL2,TOP2A CREB1 cAMP responsive element binding protein 1 6 1.93E-05 0.000174 SNAI2,SLC20A1,ODC1,MMP2,NDC80,PLAU FOS FBJ murine osteosarcoma viral oncogene homolog 5 2.61E-05 0.000209 PLAUR,MMP3,MMP1,CXCL8,PLAU JUN jun proto-oncogene 7 3.77E-05 0.000271 CXCL8,MMP2,MMP3,MMP12,PLAUR,PLAU,MMP1 2.3 Identification of core genes Using the MCODE module, we constructed a network diagram illustrating the interactions among the differentially expressed genes (Fig. 3). This network diagram provides a comprehensive depiction of the pathways associated with these genes. Referring to the STING database and conducting a joint analysis, we identified the core genes with a connectivity score of 10 or higher. The core genes include FN1, ACVR1, COL1A2, COL3A1, CDC45, CDK1, CDC6, CDKN3, CENPF, COL4A1, COL4A2, CXCL1, FEN1, FOXM1, CDH11, CKS1B, HMMR, MMP1, MMP2, KIF14, COL5A2, IGFBP3, and LUM. 2.4 Assessment of expression and correlation of core genes To depict the correlation of gene expression, we conducted an analysis using the TCGA database’s esophageal squamous cell carcinoma (ESCC) dataset. Firstly, we generated a correlation heatmap (Fig. 4A). The heatmap analysis revealed that these genes could be divided into two subgroups, exhibiting high correlation within each subgroup. One subgroup is associated with cell proliferation regulation, while the other subgroup is related to extracellular matrix-associated proteins. The mRNA expression of these genes was significantly upregulated in cancer tissues compared to adjacent non-cancerous tissues (Fig. 4B). 2.5 Development of a prognostic model based on core genes To investigate the significance of core genes in esophageal squamous cell carcinoma, we utilized the TCGA dataset, which provides more clinical information. Using the Cox algorithm, we developed a weighted risk assessment algorithm based on the core genes. The specific formula is as follows: Risk score = (0.581) * FN1 + (0.7088) * ACVR1 + (1.4492) * COL1A2 + (-1.6488) * COL3A1 + (-0.5607) * CDC45 + (0.6144) * CDK1 + (-0.6485) * CDC6 + (-0.019) * CDKN3 + (1.6349) * CENPF + (1.1485) * COL4A1 + (-0.8048) * COL4A2 + (-0.1334) * CXCL1 + (-0.7669) * FEN1 + (-0.0861) * FOXM1 + (-0.4379) * CDH11 + (1.1661) * CKS1B + (0.1452) * HMMR + (0.2036) * MMP1 + (-0.3391) * MMP2 + (-2.4686) * KIF14 + (-0.3142) * COL5A2 + (0.3651) * IGFBP3 + (0.2296) * LUM. Using this algorithm, we evaluated the risk score for each patient and ranked them from low to high (Fig. 5A). The survival status of the patients is shown in Figure 5B. The expression of core genes in each patient is illustrated in Figure 5C. After stratifying patients based on the risk score, we observed significantly worse prognosis in the high-risk group compared to the low-risk group (Fig. 5D, p=0.000245). The analysis of the true positive fraction demonstrated that our prognostic risk model exhibited excellent predictive performance (Fig. 5E, 3-year AUC=0.98). 2.6 Pan-cancer study to investigate the prognostic value of core genes. In order to investigate whether the core genes identified in ESCC are applicable to multiple cancers, we analyzed the prognostic value of the ESCC core genes in various tumors using the Cox algorithm (Table 2). Among all the cancers analyzed, ESCC and ESCA had the lowest AIC values. For ESCC, the 1-year AUC was 0.78, and the 3-year AUC was 0.98. For ESCA, the 1-year, 3-year, and 5-year AUC values were 0.908, 0.881, and 1, respectively. Furthermore, in both ESCC and ESCA, there were significant differences in median survival between the high-risk and low-risk groups when stratified based on the core genes. However, in the remaining tumors, there were no significant differences observed, unlike ESCC and ESCA. Highlighting the tumor specificity and high prognostic value of the core genes. The heatmap depicting the relationship between cancer types and the weighted coefficients of genes was shown in Figure 6. Table 2. Pan-cancer validation of gene models Cancer Log P AIC 1 Year 3 Year 5 Year Medina Time High/Low ESCC 1.40E-05 197.0698 0.78 0.98 1.5/3.5 ESCA 2,84E-6 265.38 0.908 0.881 1 1.1/4.6 LIHC 1.23E-06 1325.05 0.767 0.7 0.677 2.5/6.7 STAC 1.92E-04 1508.84 0.656 0.677 0.789 2/4.8 COAD 6.02E-05 1046.91 0.688 0.655 0.703 4.9/- GBM 1.15E-07 978.70 0.795 0.773 0.891 0.7/1.4 LGG 2.04E-11 1136.84 0.89 0.871 0.823 4.1/8.8 HNSCC 4.93E-07 2381.62 0.649 0.662 0.622 2.5/5.7 PANC 7.27E-07 2381.62 0.755 0.777 0.73 1.3/3.8 LUAD 9.14E-07 1936.33 0.676 0.682 0.67 3.1/5.9 LUSC 1.94E-03 2282.81 0.525 0.619 0.595 3.1/5.7 BRCA 6.84E-05 1758.55 0.643 0.651 0.672 9.5/11.9 CESC 1.45E-05 573.90 0.813 0.712 0.722 3.8/- SKCM 1.02E-13 2194.10 0.72 0.667 0.692 4.1/13.9 BLCA 4.51E-06 1879.76 0.701 0.67 0.64 1.8/6.6 PRAD 2.05E-05 1879.76 0.739 0.752 0.683 3.1/6.7 LAML 3.89E-04 656.29 0.728 0.755 0.77 0.7/2.2 2.7 Prognostic model and signaling pathways To investigate the impact of core gene algorithms on cell signaling pathways or processes, we analyzed the mRNA expression differences between high-risk and low-risk patient groups after evaluating the core genes. In comparison to the low-risk group, the high-risk group showed 663 genes with higher expression and 205 genes with lower expression. The volcano plot and heatmap of gene expression are shown in Figure 7A and 7B, respectively. Furthermore, through GO clustering, we found that the highly expressed genes in the high-risk group are mainly involved in focal adhesion, ECM-receptor interaction, and the PI3K-AKT signaling pathway. 2.8 The correlation between core genes and immune cells within ESCC. In order to further elucidate the mechanisms underlying the poor prognosis in high-risk group patients, we analyzed the correlation between each core gene and immune cells within tumor (Fig. 8A). The analysis results indicated a significant positive correlation between the core genes and M1 macrophages, M2 macrophages and regulatory T cells. Notably, 56% of the core genes showed a significant positive correlation with M2 macrophages. Furthermore, we constructed a network diagram illustrating the correlation between the risk coefficient and immune cells as well as the correlation between immune cells themselves (Fig. 8B). The analysis results suggested that our prognostic model exhibited the most significant correlation with the quantity of M2 macrophages in the tumor microenvironment. The network analysis also revealed a significant negative correlation between the quantity of follicular helper T cells and the quantity of M2 macrophages. 2.9 Core genes and chemotherapy resistance After analyzing the association between core genes and the prognosis of ESCC, as well as the relevant molecular mechanisms, we were interested in the potential benefits of frontline chemotherapy in patients of the high-risk group. The analysis results implied that tumors in high-risk group have higher IC50 values for cisplatin, suggesting that the benefit of standard cisplatin treatment may be compromised in high-risk group (Fig. 9A). The drug sensitivity to paclitaxel (Fig. 9B), doxorubicin (Fig. 9C), and BI.2536 (Fig. 9D) also showcased similar characteristics across high and low-risk groups. Conversely, no significant difference in drug sensitivity was observed between the two groups with respect to the frontline chemotherapy drug 5-fluorouracil (Fig. 9E). However, what surprises us is that compared to the low-risk group, patients in the high-risk group show strong sensitivity to metformin (Fig. 9F). 2.10 Core genes and immunotherapy analysis Previous analyses indicated that high-risk patients might not derive significant benefits from standard chemotherapy. Consequently, we assessed the correlation between the expression of core genes and common immune checkpoint genes (Fig. 10A). A notable positive correlation was revealed between core genes and checkpoint relative genes SIGLEC15, PDCD1LG2, and HVCR2. Further analysis also shows that tumors in high-risk groups possess lower TIDE values, suggesting that immunotherapy could potentially offer improved outcomes for patients in the high-risk group (Fig. 10B). Discussion In our investigation, we utilized multiple datasets for examination, identifying common differentially expressed genes across various datasets. Network analysis of pathway clustering suggested that among the genes we discovered, those associated with the extracellular matrix organization and regulation of cell cycle process pathways were markedly enriched (Figure 2B). These pathways are most relevant to tumor proliferation, metastasis, and prognosis and represent the most common characteristics of cancer[22]. The analysis of transcription factors suggests potential correlations between these gene expressions and transcription factors such as RELA, NFKB1, SP1, HDCA1, E2F1, HIF1, etc. RELA and NFKB1 both belong to the NFKB family, playing pivotal roles in the production of inflammatory factors and the activation of corresponding pathways[23]. Their significance often emerges in the development stages of esophageal cancer, where prolonged chronic inflammation is a usual accompaniment. This phenomenon has been validated in both murine models[24] and clinical samples[25]. Potential sources of inflammation may include overheated diet, burns to the esophageal mucosa induced by alcohol consumption[26], or settlement of pathogenic bacteria in the esophagus due to poor oral hygiene[27, 28]. Genes associated with the transcription factors RELA and NFKB1 include SERPINE1, MMP1, MMP3, and CXCL1, which are renowned for their ability to promote tumor metastasis[29]. The transcription factor SP1 also has similar target genes, however, the prognostic value of SP1 in ESCC and its role in tumor progression has not been fully researched yet. Cell proliferation transcription factors such as E2F1 are correlated with genes like SERPINE1, CDK1, FOXM1, CDC6, AURKA, NELL2, TOP2A. The genes CDK1, AURKA, and TOP2A have been demonstrated to bear significant prognostic value in various types of cancers[30, 31]. The gene interaction network presents core genes among all identified genes, which highly resemble those generated from the transcription factor analysis. This indicates that the roles of transcription factors, such as SP1, may warrant in-depth investigation in studies associated with ESCC. Subsequently, we screened core genes based on the interaction network, including KIF14, CENPF, CDK1, CDC6, CDC45, FOXM1, CKS1B, FEN1, HMMR, CDKN3, IGFBP3, CXCL1, COL4A2, COL4A1, ACVR1, COL3A1, COL1A2, COL5A2, FNI, LUM, CDH11, MMP2, MMP1. Using the TCGA-ESCC dataset as a basis, we analyzed the interrelationship of mRNA expressions among core genes (Figure 4A). These genes can be distinctively divided into two groups, namely, cell cycle and tumor EMT-related genes. The analysis of gene mRNA expression reveals that all selected genes are significantly up-regulated in tumor tissues compared to healthy esophageal tissue. Subsequently, we employed the Cox model, obtaining a prognostic model from the TCGA-ESCC dataset that can effectively predict patient outcomes. Although some studies have reported on gene-set-based prognostic models, their AUC values are not high, possibly due to their method of selectively using certain gene classes to build their prognostic models. Our model, while not incorporating some renowned genes associated with cancer prognosis like ARUKA and NEK2[32], nevertheless achieved the best result reported until now, with a 3-year AUC value of 0.98. We used the Cox algorithm to compute the gene set across multiple cancer databases. The results demonstrated that the gene set also has excellent prognostic value in ESCA, with a 5-year AUC value of 1, and its ACI value is just slightly higher than that of ESCC. This suggests that this gene set might be applicable to both ESCC and ESCA, indicating its high specificity across a wide range of cancers. Cancer cells reorganize and reprogram host cells, and regulate the tumor-supporting environment by reshaping the vascular system and extracellular matrix. This dynamic process depends on the interactions between cancer cells and the resident or recruited non-cancerous cells in the tumor microenvironment[33]. Macrophages are highly plastic cells with various functions, including tissue development and balance, clearance of cell debris, elimination of pathogens, and regulation of inflammatory responses[34]. While the range of macrophage activation states is complex, it is generally simplified into two types: CLASSICAL M1-activated macrophages or ALTERNATIVE M2-activated macrophages[35]. M1 macrophage polarization is driven by exposure to GM-CSF, IFN-γ, TNF-α, lipopolysaccharide (LPS), or other pathogen-associated molecular patterns[36]. It is crucial in the response to bacterial and viral pathogens and may possibly participate in anti-tumor immunity[37]. M2 macrophages play a crucial role in normal immune function and internal balance, such as stimulating Th2 responses, eradicating parasites, immune regulation, wound healing, and tissue regeneration. They promote tumor angiogenesis and metastasis by secreting factors such as VEGF and matrix metalloproteinases, which reshape the tumor microenvironment (TME), increase blood vessel formation, and promote tumor cell migration [38, 39]. In fact, the number of M2 macrophages gradually increases in the early stages of esophageal cancer, and it can be clearly observed even at the hyperplasia stage[40, 41]. The long-term chronic inflammation leads to an imbalance in the immune microenvironment of esophageal cancer[42], which is consistent with what we observed in our network analysis (Fig.8). Nearly half of the core genes show a significant positive correlation with the number of M2 cells, and the gene signature also has the most pronounced correlation with M2 macrophages. Recent studies also showed that M2 polarization of macrophages can enhance the expression of PD-L2 in tumor-associated macrophages, leading to immune evasion and tumor promotion through the PD-1 pathway. The blockade of the CCL2-CCR2 axis can significantly hinder the recruitment of tumor-associated macrophages, resulting in a positive response to treatment[43]. In our model, we observed a strong correlation between the core genes and PD1L2 expression (Fig.10A). In addition, we noticed another checkpoint gene, SIGLEC-15[44], which has strong tumor immune suppression ability, and it displays even stronger correlation with our core genes. This suggests that immunotherapy may yield better benefits in the treatment of high-risk patients. The prediction of ICB response based on TIDE supports our hypothesis (Fig.10B). However, when we predicted the correlation between first-line chemotherapy and the prognostic model, we found that high-risk group patients were less sensitive to most traditional drugs (Fig. 9). But, the miraculous metformin, which has multiple abilities, showed high sensitivity in the high-risk group. This could be related to several factors, such as metformin’s ability to reduce tumor metabolism, improve the tumor microenvironment, and reactivate tumor immunity. It has been reported that metformin has the ability of reverse the polarization of M2 macrophages[45]. Studies have also shown that metformin can reduce the number of myeloid-derived suppressor cells and M2 macrophages in tumors[46]. Combined treatment of metformin with anti-PD-1/PD-L1 antibodies might bring more benefits to patients[47]. Declarations Funding: This work was supported in part by grants from the Technology Commission Foundation of Shanxi Province (Grant No. 20210302124292, 20210302124296 and 20210302124091), Science and technology innovation plan of Shanxi Higher Education Institutions (Grant No. 2021L343, 2020L0373, 2021L342) and Changzhi Medical College doctoral research fund (Grant No. 521300). Data availability: The datasets used and/or analysed during the current study available from the corresponding author on reasonable request. References Abnet CC, Arnold M, Wei WQ. Epidemiology of Esophageal Squamous Cell Carcinoma. Gastroenterology. 2018; 154: 360-73. Yang YM, Hong P, Xu WW, He QY, Li B. Advances in targeted therapy for esophageal cancer. Signal Transduct Target Ther. 2020; 5: 229. McCormack VA, Menya D, Munishi MO, Dzamalala C, Gasmelseed N, Leon Roux M, Assefa M, Osano O, Watts M, Mwasamwaja AO, Mmbaga BT, Murphy G, Abnet CC, et al. Informing etiologic research priorities for squamous cell esophageal cancer in Africa: A review of setting-specific exposures to known and putative risk factors. Int J Cancer. 2017; 140: 259-71. He S, Xu J, Liu X, Zhen Y. Advances and challenges in the treatment of esophageal cancer. Acta Pharm Sin B. 2021; 11: 3379-92. Zhang BH, Yang J, Jiang L, Lyu T, Kong LX, Tan YF, Li B, Zhu YF, Xi AY, Xu X, Yan LN, Yang JY. Development and validation of a 14-gene signature for prognosis prediction in hepatocellular carcinoma. Genomics. 2020; 112: 2763-71. Lu T, Xu R, Li Q, Zhao JY, Peng B, Zhang H, Guo JD, Zhang SQ, Li HW, Wang J, Zhang LY. Systematic profiling of ferroptosis gene signatures predicts prognostic factors in esophageal squamous cell carcinoma. Mol Ther Oncolytics. 2021; 21: 134-43. Ye J, Wu Y, Cai H, Sun L, Deng W, Liang R, Han A. Development and Validation of a Ferroptosis-Related Gene Signature and Nomogram for Predicting the Prognosis of Esophageal Squamous Cell Carcinoma. Front Genet. 2021; 12: 697524. Song K, Gu B, Ma C, Wang B, Wang N, Yu R, Chen H. Epithelial-Mesenchymal Transition Gene Signature Is Associated with Neoadjuvant Chemoradiotherapy Resistance and Prognosis of Esophageal Squamous Cell Carcinoma. Dis Markers. 2022; 2022: 3534433. Shi X, Li Y, Pan S, Liu X, Ke Y, Guo W, Wang Y, Ruan Q, Zhang X, Ma H. Identification and validation of an autophagy-related gene signature for predicting prognosis in patients with esophageal squamous cell carcinoma. Sci Rep. 2022; 12: 1960. Bardou P, Mariette J, Escudie F, Djemiel C, Klopp C. jvenn: an interactive Venn diagram viewer. BMC Bioinformatics. 2014; 15: 293. Zhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, Benner C, Chanda SK. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019; 10: 1523. Hochberg Y, Benjamini Y. More powerful procedures for multiple significance testing. Stat Med. 1990; 9: 811-8. Han H, Shim H, Shin D, Shim JE, Ko Y, Shin J, Kim H, Cho A, Kim E, Lee T, Kim H, Kim K, Yang S, et al. TRRUST: a reference database of human transcriptional regulatory interactions. Sci Rep. 2015; 5: 11432. Bader GD, Hogue CW. An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinformatics. 2003; 4: 2. Yi L, Wu G, Guo L, Zou X, Huang P. Comprehensive Analysis of the PD-L1 and Immune Infiltrates of m(6)A RNA Methylation Regulators in Head and Neck Squamous Cell Carcinoma. Mol Ther Nucleic Acids. 2020; 21: 299-314. Jin K, Qiu S, Jin D, Zhou X, Zheng X, Li J, Liao X, Yang L, Wei Q. Development of prognostic signature based on immune-related genes in muscle-invasive bladder cancer: bioinformatics analysis of TCGA database. Aging (Albany NY). 2021; 13: 1859-71. Xu F, Huang X, Li Y, Chen Y, Lin L. m(6)A-related lncRNAs are potential biomarkers for predicting prognoses and immune responses in patients with LUAD. Mol Ther Nucleic Acids. 2021; 24: 780-91. Li T, Fu J, Zeng Z, Cohen D, Li J, Chen Q, Li B, Liu XS. TIMER2.0 for analysis of tumor-infiltrating immune cells. Nucleic Acids Res. 2020; 48: W509-W14. Xu L, Deng C, Pang B, Zhang X, Liu W, Liao G, Yuan H, Cheng P, Li F, Long Z, Yan M, Zhao T, Xiao Y, et al. TIP: A Web Server for Resolving Tumor Immunophenotype Profiling. Cancer Res. 2018; 78: 6575-80. Jiang Q, Sun J, Chen H, Ding C, Tang Z, Ruan Y, Liu F, Sun Y. Establishment of an Immune Cell Infiltration Score to Help Predict the Prognosis and Chemotherapy Responsiveness of Gastric Cancer Patients. Front Oncol. 2021; 11: 650673. Jiang P, Gu S, Pan D, Fu J, Sahu A, Hu X, Li Z, Traugh N, Bu X, Li B, Liu J, Freeman GJ, Brown MA, et al. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med. 2018; 24: 1550-8. Hanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov. 2022; 12: 31-46. Zinatizadeh MR, Schock B, Chalbatani GM, Zarandi PK, Jalali SA, Miri SR. The Nuclear Factor Kappa B (NF-kB) signaling in cancer development and immune diseases. Genes Dis. 2021; 8: 287-97. Li X, Ding F, Wang L, Chen H, Liu Z. Disruption of enhancer-driven S100A14 expression promotes esophageal carcinogenesis. Cancer Lett. 2022; 545: 215833. Zhu F, Willette-Brown J, Song NY, Lomada D, Song Y, Xue L, Gray Z, Zhao Z, Davis SR, Sun Z, Zhang P, Wu X, Zhan Q, et al. Autoreactive T Cells and Chronic Fungal Infection Drive Esophageal Carcinogenesis. Cell Host Microbe. 2017; 21: 478-93 e7. Wang Y, Zhu C, Wang Y, Sun J, Ling D, Wang L. Survival risk prediction model for ESCC based on relief feature selection and CNN. Comput Biol Med. 2022; 145: 105460. Nomburg J, Bullman S, Nasrollahzadeh D, Collisson EA, Abedi-Ardekani B, Akoko LO, Atkins JR, Buckle GC, Gopal S, Hu N, Kaimila B, Khoshnia M, Malekzadeh R, et al. An international report on bacterial communities in esophageal squamous cell carcinoma. Int J Cancer. 2022; 151: 1947-59. Wei J, Li R, Lu Y, Meng F, Xian B, Lai X, Lin X, Deng Y, Yang D, Zhang H, Li L, Ben X, Qiao G, et al. Salivary microbiota may predict the presence of esophageal squamous cell carcinoma. Genes Dis. 2022; 9: 1143-51. Huang Y, Hong W, Wei X. The molecular mechanisms and therapeutic strategies of EMT in tumor progression and metastasis. J Hematol Oncol. 2022; 15: 129. Deng JL, Xu YH, Wang G. Identification of Potential Crucial Genes and Key Pathways in Breast Cancer Using Bioinformatic Analysis. Front Genet. 2019; 10: 695. Mou PK, Yang EJ, Shi C, Ren G, Tao S, Shim JS. Aurora kinase A, a synthetic lethal target for precision cancer medicine. Exp Mol Med. 2021; 53: 835-47. Su W, Hu H, Ding Q, Wang M, Zhu Y, Zhang Z, Geng Z, Lin S, Zhou P. NEK2 promotes the migration and proliferation of ESCC via stabilization of YAP1 by phosphorylation at Thr-143. Cell Commun Signal. 2022; 20: 87. DeNardo DG, Ruffell B. Macrophages as regulators of tumour immunity and immunotherapy. Nat Rev Immunol. 2019; 19: 369-82. Varol C, Mildner A, Jung S. Macrophages: development and tissue specialization. Annu Rev Immunol. 2015; 33: 643-75. Mosser DM, Edwards JP. Exploring the full spectrum of macrophage activation. Nat Rev Immunol. 2008; 8: 958-69. Jaguin M, Houlbert N, Fardel O, Lecureur V. Polarization profiles of human M-CSF-generated macrophages and comparison of M1-markers in classically activated macrophages from GM-CSF and M-CSF origin. Cell Immunol. 2013; 281: 51-61. Italiani P, Boraschi D. From Monocytes to M1/M2 Macrophages: Phenotypical vs. Functional Differentiation. Front Immunol. 2014; 5: 514. Yang L, Zhang Y. Tumor-associated macrophages: from basic research to clinical application. J Hematol Oncol. 2017; 10: 58. Zhang X, Ji L, Li MO. Control of tumor-associated macrophage responses by nutrient acquisition and metabolism. Immunity. 2023; 56: 14-31. Jia Y, Zhang B, Zhang C, Kwong DL, Chang Z, Li S, Wang Z, Han H, Li J, Zhong Y, Sui X, Fu L, Guan X, et al. Single-Cell Transcriptomic Analysis of Primary and Metastatic Tumor Ecosystems in Esophageal Squamous Cell Carcinoma. Adv Sci (Weinh). 2023; 10: e2204565. Dinh HQ, Pan F, Wang G, Huang QF, Olingy CE, Wu ZY, Wang SH, Xu X, Xu XE, He JZ, Yang Q, Orsulic S, Haro M, et al. Integrated single-cell transcriptome analysis reveals heterogeneity of esophageal squamous cell carcinoma microenvironment. Nat Commun. 2021; 12: 7335. Ye H, Li X, Lin J, Yang P, Su M. CD98hc has a pivotal role in maintaining the immuno-barrier integrity of basal layer cells in esophageal epithelium. Cancer Cell Int. 2022; 22: 98. Yang H, Zhang Q, Xu M, Wang L, Chen X, Feng Y, Li Y, Zhang X, Cui W, Jia X. CCL2-CCR2 axis recruits tumor associated macrophages to induce immune evasion through PD-1 signaling in esophageal carcinogenesis. Mol Cancer. 2020; 19: 41. Wang J, Sun J, Liu LN, Flies DB, Nie X, Toki M, Zhang J, Song C, Zarr M, Zhou X, Han X, Archer KA, O'Neill T, et al. Siglec-15 as an immune suppressor and potential target for normalization cancer immunotherapy. Nat Med. 2019; 25: 656-66. de Oliveira S, Houseright RA, Graves AL, Golenberg N, Korte BG, Miskolci V, Huttenlocher A. Metformin modulates innate immune-mediated inflammation and early progression of NAFLD-associated hepatocellular carcinoma in zebrafish. J Hepatol. 2019; 70: 710-21. Kang J, Lee D, Lee KJ, Yoon JE, Kwon JH, Seo Y, Kim J, Chang SY, Park J, Kang EA, Park SJ, Park JJ, Cheon JH, et al. Tumor-Suppressive Effect of Metformin via the Regulation of M2 Macrophages and Myeloid-Derived Suppressor Cells in the Tumor Microenvironment of Colorectal Cancer. Cancers (Basel). 2022; 14. Wei Z, Zhang X, Yong T, Bie N, Zhan G, Li X, Liang Q, Li J, Yu J, Huang G, Yan Y, Zhang Z, Zhang B, et al. Boosting anti-PD-1 therapy with metformin-loaded macrophage-derived microparticles. Nat Commun. 2021; 12: 440. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 14 Apr, 2024 Read the published version in Cancer Investigation → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3208103","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":222907902,"identity":"fc2c7f1b-b8d1-4c66-87ef-437622fb6090","order_by":0,"name":"Pu Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYBACefnHx39IVPyT42dvIFKLYUNagoTFmQPGkj0HiLXmQI6BRGXLgcQNNxKI1MHYcMDA4GbDHcYNNx9vvMFQYxNNUAs7Y0NC4swdz5glb6cVWzAcS8ttIGhLM8OBw5JnmNn4bueYSTA2HCasheEYY2Pz3zZmHoabZ4jVcoaZmUGy7bCEwA0eIrUYzmBjY5A4k2Yg2QP0SwIxfpGX4P/GIFFhU9/PfnjjjQ81NkQ4DAkYSCSQohyihVQdo2AUjIJRMDIAAM5yQ31Xd7UJAAAAAElFTkSuQmCC","orcid":"","institution":"Changzhi Medical College","correspondingAuthor":true,"prefix":"","firstName":"Pu","middleName":"","lastName":"Wang","suffix":""},{"id":222907903,"identity":"bf18d9e1-5183-4a2f-971d-43438e8aaa7d","order_by":1,"name":"Bin Du","email":"","orcid":"","institution":"Changzhi Medical College","correspondingAuthor":false,"prefix":"","firstName":"Bin","middleName":"","lastName":"Du","suffix":""},{"id":222907904,"identity":"20b9ce56-350a-4095-88d3-17c33d0f972b","order_by":2,"name":"Lingyu Wei","email":"","orcid":"","institution":"Heping Hospital Afliated to Changzhi Medical College","correspondingAuthor":false,"prefix":"","firstName":"Lingyu","middleName":"","lastName":"Wei","suffix":""},{"id":222907905,"identity":"c39ed98f-e6d6-4b34-8a65-46ee88799fa7","order_by":3,"name":"Jia Wang","email":"","orcid":"","institution":"Changzhi Medical College","correspondingAuthor":false,"prefix":"","firstName":"Jia","middleName":"","lastName":"Wang","suffix":""},{"id":222907906,"identity":"6bfd8550-d3a5-42e8-a302-5f4ad2b9d866","order_by":4,"name":"Jinshang Wang","email":"","orcid":"","institution":"Changzhi Medical College","correspondingAuthor":false,"prefix":"","firstName":"Jinshang","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2023-07-27 03:14:20","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3208103/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3208103/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1080/07357907.2024.2340576","type":"published","date":"2024-04-14T04:13:19+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":41097866,"identity":"4bb9f5e2-0c02-40fd-b09f-b413db56b85e","added_by":"auto","created_at":"2023-08-04 23:36:34","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":55840,"visible":true,"origin":"","legend":"\u003cp\u003eThe dataset used for bioinformatics analysis and Venn diagram analysis.\u003c/p\u003e\n\u003cp\u003e(A) The datasets used and the heatmap of differentially expressed genes. (B) Venn diagram analysis of differentially expressed genes in different datasets.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/72628369d2ef0c5da2ee1152.jpg"},{"id":41097867,"identity":"c2c3cccd-7645-45f3-a445-479458c2be27","added_by":"auto","created_at":"2023-08-04 23:36:34","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":84795,"visible":true,"origin":"","legend":"\u003cp\u003eBioinformatics analysis of commonly differentially expressed genes. (A) Pathway and process analysis. (B) Network analysis of pathways and processes. (C) Transcription factor analysis using TRRUST database.\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/a107e96489b89dd831fe1578.jpg"},{"id":41097872,"identity":"1fa1c5b6-3d20-4d5e-b02d-20f0007c1ddb","added_by":"auto","created_at":"2023-08-04 23:36:35","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":112538,"visible":true,"origin":"","legend":"\u003cp\u003eProtein-protein interaction network analysis of commonly differentially expressed genes.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/01dd3b85bf13d132889cce95.jpg"},{"id":41099859,"identity":"f1d829be-b171-4226-a418-3292ba37eac4","added_by":"auto","created_at":"2023-08-04 23:52:35","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":117508,"visible":true,"origin":"","legend":"\u003cp\u003eExpression and interrelationships of core genes in tumors. (A) Heatmap of the correlation between the expression of core genes, the expression correlation of genes was analysed with Spearman; (B) Expression of core genes in adjacent non-cancerous tissues and ESCC tissues.\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/cc6db2a13787601069678f4c.jpg"},{"id":41099598,"identity":"384b0d09-3856-47a1-a48f-9f3fe1afdb17","added_by":"auto","created_at":"2023-08-04 23:44:35","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":95971,"visible":true,"origin":"","legend":"\u003cp\u003eEstablishment of a prognostic model based on core genes. (A) Risk score and grouping of patients; (B) Survival status of patients; (C) Heatmap of the scores of each core gene for each sample; (D) Analysis of high and low-risk patients and their prognosis; (E) Kaplan-Meier curves of the prognostic model at 1 and 3 years.\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/4d7a7209c4dbe01da9c771e4.jpg"},{"id":41099597,"identity":"daf01885-a900-4be0-9a42-95b24a305b57","added_by":"auto","created_at":"2023-08-04 23:44:35","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":56929,"visible":true,"origin":"","legend":"\u003cp\u003eHeatmap of the scores of each core gene in different tumors in our prognostic model.\u003c/p\u003e","description":"","filename":"6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/81bf7a216c78e172dd46f9cd.jpg"},{"id":41101625,"identity":"e69c5fc3-0f18-4a21-af19-e60e1c0a9709","added_by":"auto","created_at":"2023-08-05 00:00:35","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":137321,"visible":true,"origin":"","legend":"\u003cp\u003eSignaling pathways associated with the prognostic model. (A) Volcano plot of differentially expressed genes; (B) Heatmap of differentially expressed genes; (C) KEGG clustering analysis of downregulated genes in the high-risk group compared to the low-risk group; (D) KEGG clustering analysis of upregulated genes in the high-risk group compared to the low-risk group.\u003c/p\u003e","description":"","filename":"7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/a71016ac47823974a3533cd9.jpg"},{"id":41097875,"identity":"470f7881-fe9b-4734-b116-bf63f3b56a44","added_by":"auto","created_at":"2023-08-04 23:36:35","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":273139,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelation between prognostic model and immune cells in tumors. (A)Heatmap of the correlation between multiple genes or models and immune score.The abscissa and ordinate represent genes, different colors represent different correlation coefficients, blue represents positive correlation whereas red represents negative correlation, the darker the color, the stronger the relation; (B) The heat map in the schematic represents the correlation analysis of the immune score itself, red represents positive correlation, blue represents negative correlation, the more red or blue color means the greater correlation, also the larger circle means the stronger correlation; the red line in the schematic represents the negative correlation between the model score or gene expression and the immune score, green means the positive correlation. ** for p \u0026lt; 0.01, * for p \u0026lt; 0.05.\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/379064fdb40aac5384c8161f.png"},{"id":41097874,"identity":"85d46be8-2cda-4fd3-a755-5114d5b72946","added_by":"auto","created_at":"2023-08-04 23:36:35","extension":"jpg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":98017,"visible":true,"origin":"","legend":"\u003cp\u003ePrediction of chemotherapy drug sensitivity in high and low-risk groups of patients. The prognostic model predicts patients’ sensitivity to (A) cisplatin, (B) paclitaxel, (C) 5-fluorouracil, (D) doxorubicin, (E) BI2536, and (F) metformin. *p \u0026lt; 0.05, **p \u0026lt; 0.01, ***p \u0026lt; 0.001\u003c/p\u003e","description":"","filename":"9.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/69181140f0961aa33c1077b2.jpg"},{"id":41097869,"identity":"7a41a3fc-7f12-4dde-879a-645b6b2009ee","added_by":"auto","created_at":"2023-08-04 23:36:35","extension":"jpg","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":57984,"visible":true,"origin":"","legend":"\u003cp\u003ePrediction of prognostic model and immune therapy response. (A) Correlation between core gene expression and immune checkpoint gene expression. (B) Prediction of therapeutic effect of immune checkpoint blockade therapy in high-risk group patients and low-risk group patients. **** p\u0026lt;0.0001\u003c/p\u003e","description":"","filename":"10.jpg","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/335e3c300cdafa25af03cccb.jpg"},{"id":54668801,"identity":"29db7e2c-d1bd-418d-9fcb-c9d2cac60c71","added_by":"auto","created_at":"2024-04-15 04:21:34","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1463970,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3208103/v1/bad3fa04-95a1-4519-814b-f4439fc2eb85.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Development of a Novel Prognostic Model for Esophageal Squamous Cell Carcinoma: Insights into Immune Cell Interactions and Drug Sensitivity","fulltext":[{"header":"Introduction","content":"\u003cp\u003eEsophageal cancer (EC) is ranked as the eighth most common cancer and the sixth leading cause of cancer-associated mortality, with a low five-year overall survival rate[1]. Base on pathological characteristics, EC is typically categorized into Esophageal Squamous Cell Carcinoma (ESCC) and Esophageal Adenocarcinoma (EAC). The characteristics and causes of EC may vary by regions or races. In China, ESCC accounts for 90-95% of all esophageal cancer incidences[2]. Smoking, alcohol consumption, consumption of overheated water, and other environmental factors are recognized as key risk elements for esophageal squamous cell carcinoma[3]. Recent advancements in surgical techniques and the use of neoadjuvant chemotherapy have significantly improved the prognosis for patients with esophageal squamous cell carcinoma. However, as most patients with esophageal cancer are diagnosed in the late stages, the five-year survival rate for patients with esophageal squamous cell carcinoma still remains less than 20%[4]. Therefore, the development of molecular markers that can guide the prognosis and treatment plan for ESCC patients is of great clinical significance.\u003c/p\u003e\n\u003cp\u003eThe advancement of bioinformatics in recent years has facilitated the development of gene signatures based on the expression of multiple genes, offering new directions for molecular subtyping and treatment of tumors[5]. However, this field of study has not seen significant progress in esophageal cancer, largely due to insufficiencies in data sets. In existing studies, researchers have utilized multiple datasets and examined genes associated with signal pathways, including ferroptosis [6, 7] , epithelial-mesenchymal transition [8], and autophagy-related genes [9]. Nevertheless, due to various factors such as lack of data sets, inferior quality of data sets, and incompleteness of research schemes, the established gene signatures have not shown substantive discriminative capability in prognosticating for ESCC patients, with AUC values often distributed around 0.7. Consequently, the currently established gene signatures remain less practicable.\u003c/p\u003e\n\u003cp\u003eIn our study, we utilized multiple ESCC datasets encompassing paired adjacent non-cancerous and cancerous tissues. We analyzed their common differential genes, and based on this, we established a multi-gene marker composed of core genes, which we subsequently optimized. Finally, through a wide variety of analyses, we unraveled the inherent molecular mechanisms through which these gene markers can efficiently predict the prognosis of ESCC patients.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cstrong\u003e1.1 Data acquisition\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe available datasets related to esophageal squamous cell carcinoma (ESCC) are mainly obtained from the Gene Expression Omnibus (GEO, www.ncbi.nlm.nih.gov/geo/) and The Cancer Genome Atlas (TCGA, www.cancer.gov/) databases. The GSE22954, GSE77861, and GSE20347 datasets are obtained from the GEO database. The ESCC dataset with prognosis information is sourced from the TCGA database. The inclusion criteria for data sets are: (a)having a sufficient number of esophageal squamous cell carcinoma samples; (b) having paired adjacent cancer tissue or healthy tissue samples.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.2 Acquisition of Differentially Expressed Genes \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe obtained RNA-sequencing expression (level 3) profiles and associated clinical data for ESCC from the GEO and TCGA datasets. The limma package in R software was used to analyze the differentially expressed mRNA. The threshold for differential expression of mRNAs was defined as \u0026ldquo;Adjusted P \u0026lt; 0.05 and Log2 (Fold Change) \u0026gt;1 or Log2(Fold Change)\u0026lt; \u0026minus;1.\u0026rdquo;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.3 Analysis of shared differentially expressed genes\u003c/p\u003e\n\u003cp\u003eTo the identification of shared differentially expressed genes across multiple gene sets, we employed VEN analysis. The construction of Venn diagrams was performed using the JVENN tool [10]. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.4 Pathway and process enrichment analysis\u003c/p\u003e\n\u003cp\u003ePathway and process enrichment analysis were performed at Metascape (metascape.org) [11]. The enrichment analysis for each pathway and process was executed using given gene lists and the following ontology sources: KEGG Pathway, GO Biological Processes, Reactome Gene Sets, Canonical Pathways, CORUM, WikiPathways, and PANTHER Pathway. The enrichment background used all genes in the genome. Terms gathered included those with a p-value of below 0.01, a minimum of 3 counts, and an enrichment factor above 1.5, which is the ratio of the observed counts to the expected counts predicted by chance. These terms were then organized into clusters based on their mutual traits. P-values were derived from the cumulative hypergeometric distribution, whilst q-values were determined using the Benjamini-Hochberg procedure to account for any multiple testings[12]. The similarity metric for the hierarchical clustering of the enriched terms applied Kappa scores, considering sub-trees with a similarity above 0.3 as a cluster. The most statistically significant term within a cluster was selected to represent that cluster.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.5 Pathways network analysis\u003c/p\u003e\n\u003cp\u003eTo better understand the links between the terms, we\u0026rsquo;ve chosen a subset of enriched terms and displayed them in a network plot. Here, terms with a similarity of over 0.3 are linked by edges. From each of the 20 clusters, we\u0026rsquo;ve specifically chosen the terms with the top p-values. However, we\u0026rsquo;ve ensured that each cluster doesn\u0026rsquo;t exceed 15 terms and the total term count doesn\u0026rsquo;t surpass 250. The network visualization was achieved using Cytoscape.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.6 Genes relative transcription factor analysis\u003c/p\u003e\n\u003cp\u003eThe transcription factor analysis related to differentially expressed genes was performed using TRRUST online analysis [13].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.7 Protein-protein interaction enrichment analysis\u003c/p\u003e\n\u003cp\u003eFor each given genes, the Molecular Complex Detection (MCODE) algorithm10 were applied to identify densely connected network components[14].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.8 Correlation analysis of genes.\u003c/p\u003e\n\u003cp\u003eThe genes correlation map is realized by the R software package ggstatsplot, and the multi-gene correlation pheatmap is displayed by the R software package[15]. \u0026nbsp; Spearman\u0026rsquo;s correlation analysis to describe the correlation between quantitative variables without a normal distribution. P values less than 0.05 were considered statistically significant (*P\u0026thinsp;\u0026lt;\u0026thinsp;0.05).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.9 mRNA expression of core genes\u003c/p\u003e\n\u003cp\u003eRNA-sequencing expression (level 3) profiles and corresponding clinical information for ESCC were downloaded from the TCGA dataset(https://portal.gdc.com). All the analysis methods and R package were implemented by R version 4.0.3[16].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.10 Establishment of a prognostic model.\u003c/p\u003e\n\u003cp\u003eRNA-sequencing expression (level 3) profiles and corresponding clinical information for xx were downloaded from the TCGA dataset(https://portal.gdc.com). Converting counts data to TPM and normalizing the data log2 (TPM+1), keeping samples with clinical information at the same time. Indicted samples for subsequent analysis. Log-rank test was used to compare differences in survival between these groups. The time-ROC (v 0.4) analysis was used to compare the predictive accuracy of genes and risk score. Multivariate cox regression analysis was used to construct a prognostic model, and the R package survival used for the analysis [17].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.11 Analysis of the correlation between gene expression and the amount of immune cells in the tumor.\u003c/p\u003e\n\u003cp\u003eTRNA-sequencing expression (level 3) profiles and corresponding clinical information for ESCC were downloaded from the TCGA dataset (https://portal.gdc.com). The R software GGSTATSPLOT package was used to draw the correlations between gene expression and immune score [18], the R software heatmap package was used to draw multi-gene correlation. Used Spearman\u0026rsquo;s correlation analysis to describe the correlation between quantitative variables without a normal distribution. P values less than 0.05 were considered statistically significant (*P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e1.12 Network analysis of the correlation between the prognostic model and immune cells.\u003c/p\u003e\n\u003cp\u003eTo assess the reliable results of immune score evaluation, we used immuneeconv. It\u0026rsquo;s an R software package that integrates six latest algorithms, including TIMER, xCell, MCP-counter, CIBERSORT, EPIC and quanTIseq. These algorithms had beeen benchmarked, each had a unique advantage. Analysis and visualization are performed by the R software ggClusterNet package (https://doi.org/10.1002/imt2.32)[19].\u003c/p\u003e\n\u003cp\u003e1.13 Prediction of drug sensitivity.\u003c/p\u003e\n\u003cp\u003eRNA-sequencing expression (level 3) profiles and corresponding clinical information for ESCC were downloaded from the TCGA dataset (https://portal.gdc.com).Predicted the chemotherapeutic response for each sample based on the largest publicly available pharmacogenomics database the Genomics of Drug Sensitivity in Cancer (GDSC), https://www.cancerrxgene.org/.The prediction process was implemented by R package \u0026ldquo;pRRophetic\u0026rdquo;. The samples\u0026apos; half-maximal inhibitory concentration (IC50) was estimated by ridge regression. All parameters were set as the default values using the the batch effect of combat and tissue type of all tissues, and the duplicate gene expression was summarized as mean value[20].\u003c/p\u003e\n\u003cp\u003e1.14 Prediction of patients\u0026rsquo; sensitivity to immune checkpoint blockade(ICB) therapy.\u003c/p\u003e\n\u003cp\u003ePotential ICB response was predicted with TIDE algorithm[21].\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e2.1 Identification of signature differentially expressed genes in esophageal squamous cell carcinoma\u003c/p\u003e\n\u003cp\u003eEsophageal cancer has a very poor prognosis, but there is currently no effective prognostic evaluation model in clinical practice. We attempted to identify genes with common differential expression in esophageal squamous cell carcinoma through data analysis. Four datasets with paired cancer adjacent tissues were retrieved from the GEO and TCGA databases, namely GSE22954, GSE77861, GSE20347 (GEO), and the esophageal squamous cell carcinoma dataset in the TCGA database (Fig. 1A). Through VEN analysis, we identified 119 differentially expressed genes with high homogeneity (Fig. 1B).\u003c/p\u003e\n\u003cp\u003e2.2 Cluster analysis of differentially expressed genes\u003c/p\u003e\n\u003cp\u003eFurthermore, we performed cluster analysis on the 119 genes that showed significant differential expression in different esophageal squamous cell carcinoma databases using the Metascape website. In the pathway cluster analysis, the top five pathways were extracellular matrix organization, burn wound healing, PID integrin1 pathway, regulation of cell cycle process, and network map of SARS-CoV-2 signaling pathway (Fig. 2A). The network analysis of pathway clustering revealed that extracellular matrix organization, burn wound healing, and regulation of cell cycle process pathways predominated among the pathways showing significant clustering (Fig. 2B). Additionally, using the TRRUST module, we analyzed the transcription factors behind these differentially expressed genes. The results showed that the classical transcription factors RELA, NFKB1, and SP1 may play important roles in the expression of these genes (Fig. 2C). The relationship between the transcription factors and the corresponding overlapped genes was listed in Table 1.\u003c/p\u003e\n\u003cp\u003eTable 1. Transcription factors associated with overlapping genes\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"548\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eKey TF\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003eDescription\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003eoverlapped genes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003eP value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003eQ value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eList of overlapped genes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eRELA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003ev-rel reticuloendotheliosis viral oncogene homolog A (avian)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e14\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e4.74E-09\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e3.41E-07\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCXCL8,COL1A2,MMP2,TNFAIP3,CCL20,PLAU,MMP12\u003c/strong\u003e\u003cstrong\u003e,\u003c/strong\u003e\u003cstrong\u003eTIMP1,PLAUR,SERPINE1,MMP1,MMP3,FN1,CXCL1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eNFKB1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003enuclear factor of kappa light polypeptide gene enhancer in B-cells 1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e13\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e4.52E-08\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.63E-06\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMMP3,TNFAIP3,MMP1,TIMP1,COL1A2,CXCL8,PLAU\u003c/strong\u003e\u003cstrong\u003e,\u003c/strong\u003e\u003cstrong\u003eMMP2,FN1,PLAUR,CCL20,SERPINE1,CXCL1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSP1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSp1 transcription factor\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e14\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.20E-06\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2.55E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCDH3,HMGA2,MMP2,FOXM1,KRT14,PLAU,SERPINE1\u003c/strong\u003e\u003cstrong\u003e,\u003c/strong\u003e\u003cstrong\u003eTIMP1,CXCL1,HOXB7,ODC1,SPP1,IGFBP3,PLAUR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eETS2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003ev-ets erythroblastosis virus E26 oncogene homolog 2 (avian)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.42E-06\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2.55E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMMP1,CXCL8,MMP3,CDK1,MMP2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eHDAC1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003ehistone deacetylase 1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e4.89E-06\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e7.04E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSNAI2,IGFBP3,SPP1,CXCL8,BAMBI,COL1A2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eFLI1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003eFriend leukemia virus integration 1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.31E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.000157\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMMP1,COL1A2,IGFBP3,FOXM1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eE2F1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003eE2F transcription factor 1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e7\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.90E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.000174\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSERPINE1,CDK1,FOXM1,CDC6,AURKA,NELL2,TOP2A\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCREB1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003ecAMP responsive element binding protein 1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.93E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.000174\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSNAI2,SLC20A1,ODC1,MMP2,NDC80,PLAU\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eFOS\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003eFBJ murine osteosarcoma viral oncogene homolog\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2.61E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.000209\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePLAUR,MMP3,MMP1,CXCL8,PLAU\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"8.75912408759124%\"\u003e\n \u003cp\u003e\u003cstrong\u003eJUN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"23.722627737226276%\"\u003e\n \u003cp\u003e\u003cstrong\u003ejun proto-oncogene\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"10.766423357664234%\"\u003e\n \u003cp\u003e\u003cstrong\u003e7\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"12.043795620437956%\"\u003e\n \u003cp\u003e\u003cstrong\u003e3.77E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"8.576642335766424%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.000271\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"36.13138686131387%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCXCL8,MMP2,MMP3,MMP12,PLAUR,PLAU,MMP1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e2.3 Identification of core genes\u003c/p\u003e\n\u003cp\u003eUsing the MCODE module, we constructed a network diagram illustrating the interactions among the differentially expressed genes (Fig. 3). This network diagram provides a comprehensive depiction of the pathways associated with these genes. Referring to the STING database and conducting a joint analysis, we identified the core genes with a connectivity score of 10 or higher. The core genes include FN1, ACVR1, COL1A2, COL3A1, CDC45, CDK1, CDC6, CDKN3, CENPF, COL4A1, COL4A2, CXCL1, FEN1, FOXM1, CDH11, CKS1B, HMMR, MMP1, MMP2, KIF14, COL5A2, IGFBP3, and LUM.\u003c/p\u003e\n\u003cp\u003e2.4 Assessment of expression and correlation of core genes\u003c/p\u003e\n\u003cp\u003eTo depict the correlation of gene expression, we conducted an analysis using the TCGA database\u0026rsquo;s esophageal squamous cell carcinoma (ESCC) dataset. Firstly, we generated a correlation heatmap (Fig. 4A). The heatmap analysis revealed that these genes could be divided into two subgroups, exhibiting high correlation within each subgroup. One subgroup is associated with cell proliferation regulation, while the other subgroup is related to extracellular matrix-associated proteins. The mRNA expression of these genes was significantly upregulated in cancer tissues compared to adjacent non-cancerous tissues (Fig. 4B).\u003c/p\u003e\n\u003cp\u003e2.5 Development of a prognostic model based on core genes\u003c/p\u003e\n\u003cp\u003eTo investigate the significance of core genes in esophageal squamous cell carcinoma, we utilized the TCGA dataset, which provides more clinical information. Using the Cox algorithm, we developed a weighted risk assessment algorithm based on the core genes. The specific formula is as follows: Risk score = (0.581) * FN1 + (0.7088) * ACVR1 + (1.4492) * COL1A2 + (-1.6488) * COL3A1 + (-0.5607) * CDC45 + (0.6144) * CDK1 + (-0.6485) * CDC6 + (-0.019) * CDKN3 + (1.6349) * CENPF + (1.1485) * COL4A1 + (-0.8048) * COL4A2 + (-0.1334) * CXCL1 + (-0.7669) * FEN1 + (-0.0861) * FOXM1 + (-0.4379) * CDH11 + (1.1661) * CKS1B + (0.1452) * HMMR + (0.2036) * MMP1 + (-0.3391) * MMP2 + (-2.4686) * KIF14 + (-0.3142) * COL5A2 + (0.3651) * IGFBP3 + (0.2296) * LUM. Using this algorithm, we evaluated the risk score for each patient and ranked them from low to high (Fig. 5A). The survival status of the patients is shown in Figure 5B. The expression of core genes in each patient is illustrated in Figure 5C. After stratifying patients based on the risk score, we observed significantly worse prognosis in the high-risk group compared to the low-risk group (Fig. 5D, p=0.000245). The analysis of the true positive fraction demonstrated that our prognostic risk model exhibited excellent predictive performance (Fig. 5E, 3-year AUC=0.98).\u003c/p\u003e\n\u003cp\u003e2.6 Pan-cancer study to investigate the prognostic value of core genes.\u003c/p\u003e\n\u003cp\u003eIn order to investigate whether the core genes identified in ESCC are applicable to multiple cancers, we analyzed the prognostic value of the ESCC core genes in various tumors using the Cox algorithm (Table 2). Among all the cancers analyzed, ESCC and ESCA had the lowest AIC values. For ESCC, the 1-year AUC was 0.78, and the 3-year AUC was 0.98. For ESCA, the 1-year, 3-year, and 5-year AUC values were 0.908, 0.881, and 1, respectively. Furthermore, in both ESCC and ESCA, there were significant differences in median survival between the high-risk and low-risk groups when stratified based on the core genes. However, in the remaining tumors, there were no significant differences observed, unlike ESCC and ESCA. Highlighting the tumor specificity and high prognostic value of the core genes. The heatmap depicting the relationship between cancer types and the weighted coefficients of genes was shown in Figure 6.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTable 2. Pan-cancer validation of gene models\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"558\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCancer\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLog P\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAIC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1 Year\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e3 Year\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e5 Year\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMedina Time\u003cbr\u003e\u0026nbsp;High/Low\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eESCC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.40E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e197.0698\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.78\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.98\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.5/3.5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eESCA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2,84E-6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e265.38\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.908\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.881\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.1/4.6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLIHC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.23E-06\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1325.05\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.767\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.7\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.677\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2.5/6.7\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSTAC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.92E-04\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1508.84\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.656\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.677\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.789\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2/4.8\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCOAD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e6.02E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1046.91\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.688\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.655\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.703\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e4.9/-\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGBM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.15E-07\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e978.70\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.795\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.773\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.891\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.7/1.4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLGG\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2.04E-11\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1136.84\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.89\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.871\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.823\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e4.1/8.8\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eHNSCC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e4.93E-07\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2381.62\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.649\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.662\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.622\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2.5/5.7\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePANC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e7.27E-07\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2381.62\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.755\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.777\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.73\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.3/3.8\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLUAD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e9.14E-07\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1936.33\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.676\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.682\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.67\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e3.1/5.9\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLUSC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.94E-03\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2282.81\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.525\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.619\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.595\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e3.1/5.7\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eBRCA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e6.84E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1758.55\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.643\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.651\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.672\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e9.5/11.9\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCESC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.45E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e573.90\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.813\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.712\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.722\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e3.8/-\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSKCM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.02E-13\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2194.10\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.72\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.667\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.692\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e4.1/13.9\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eBLCA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e4.51E-06\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1879.76\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.701\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.67\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.64\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1.8/6.6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePRAD\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e2.05E-05\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e1879.76\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.739\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.752\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.683\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e3.1/6.7\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLAML\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"13.799283154121865%\"\u003e\n \u003cp\u003e\u003cstrong\u003e3.89E-04\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"19.892473118279568%\"\u003e\n \u003cp\u003e\u003cstrong\u003e656.29\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.728\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.755\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"11.827956989247312%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.77\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.025089605734767%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.7/2.2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e2.7 Prognostic model and signaling pathways\u003c/p\u003e\n\u003cp\u003eTo investigate the impact of core gene algorithms on cell signaling pathways or processes, we analyzed the mRNA expression differences between high-risk and low-risk patient groups after evaluating the core genes. In comparison to the low-risk group, the high-risk group showed 663 genes with higher expression and 205 genes with lower expression. The volcano plot and heatmap of gene expression are shown in Figure 7A and 7B, respectively. Furthermore, through GO clustering, we found that the highly expressed genes in the high-risk group are mainly involved in focal adhesion, ECM-receptor interaction, and the PI3K-AKT signaling pathway.\u003c/p\u003e\n\u003cp\u003e2.8 The correlation between core genes and immune cells within ESCC.\u003c/p\u003e\n\u003cp\u003eIn order to further elucidate the mechanisms underlying the poor prognosis in high-risk group patients, we analyzed the correlation between each core gene and immune cells within tumor (Fig. 8A). The analysis results indicated a significant positive correlation between the core genes and M1 macrophages, M2 macrophages and regulatory T cells. Notably, 56% of the core genes showed a significant positive correlation with M2 macrophages. Furthermore, we constructed a network diagram illustrating the correlation between the risk coefficient and immune cells as well as the correlation between immune cells themselves (Fig. 8B). The analysis results suggested that our prognostic model exhibited the most significant correlation with the quantity of M2 macrophages in the tumor microenvironment. The network analysis also revealed a significant negative correlation between the quantity of follicular helper T cells and the quantity of M2 macrophages.\u003c/p\u003e\n\u003cp\u003e2.9 Core genes and chemotherapy resistance\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;After analyzing the association between core genes and the prognosis of ESCC, as well as the relevant molecular mechanisms, we were interested in the potential benefits of frontline chemotherapy in patients of the high-risk group. The analysis results implied that tumors in high-risk group have higher IC50 values for cisplatin, suggesting that the benefit of standard cisplatin treatment may be compromised in high-risk group (Fig. 9A). The drug sensitivity to paclitaxel (Fig. 9B), doxorubicin (Fig. 9C), and BI.2536 (Fig. 9D) also showcased similar characteristics across high and low-risk groups. Conversely, no significant difference in drug sensitivity was observed between the two groups with respect to the frontline chemotherapy drug 5-fluorouracil (Fig. 9E). However, what surprises us is that compared to the low-risk group, patients in the high-risk group show strong sensitivity to metformin (Fig. 9F).\u003c/p\u003e\n\u003cp\u003e2.10 Core genes and immunotherapy analysis\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePrevious analyses indicated that high-risk patients might not derive significant benefits from standard chemotherapy. Consequently, we assessed the correlation between the expression of core genes and common immune checkpoint genes (Fig. 10A). A notable positive correlation was revealed between core genes and checkpoint relative genes SIGLEC15, PDCD1LG2, and HVCR2. Further analysis also shows that tumors in high-risk groups possess lower TIDE values, suggesting that immunotherapy could potentially offer improved outcomes for patients in the high-risk group (Fig. 10B).\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn our investigation, we utilized multiple datasets for examination, identifying common differentially expressed genes across various datasets. Network analysis of pathway clustering suggested that among the genes we discovered, those associated with the extracellular matrix organization and regulation of cell cycle process pathways were markedly enriched (Figure 2B). These pathways are most relevant to tumor proliferation, metastasis, and prognosis and represent the most common characteristics of cancer[22].\u003c/p\u003e\n\u003cp\u003eThe analysis of transcription factors suggests potential correlations between these gene expressions and transcription factors such as RELA, NFKB1, SP1, HDCA1, E2F1, HIF1, etc. RELA and NFKB1 both belong to the NFKB family, playing pivotal roles in the production of inflammatory factors and the activation of corresponding pathways[23]. Their significance often emerges in the development stages of esophageal cancer, where prolonged chronic inflammation is a usual accompaniment. This phenomenon has been validated in both murine models[24]\u0026nbsp;and clinical samples[25].\u003c/p\u003e\n\u003cp\u003ePotential sources of inflammation may include overheated diet, burns to the esophageal mucosa induced by alcohol consumption[26], or settlement of pathogenic bacteria in the esophagus due to poor oral hygiene[27, 28]. Genes associated with the transcription factors RELA and NFKB1 include SERPINE1, MMP1, MMP3, and CXCL1, which are renowned for their ability to promote tumor metastasis[29]. The transcription factor SP1 also has similar target genes, however, the prognostic value of SP1 in ESCC and its role in tumor progression has not been fully researched yet. Cell proliferation transcription factors such as E2F1 are correlated with genes like SERPINE1, CDK1, FOXM1, CDC6, AURKA, NELL2, TOP2A.\u003c/p\u003e\n\u003cp\u003eThe genes CDK1, AURKA, and TOP2A have been demonstrated to bear significant prognostic value in various types of cancers[30, 31]. The gene interaction network presents core genes among all identified genes, which highly resemble those generated from the transcription factor analysis. This indicates that the roles of transcription factors, such as SP1, may warrant in-depth investigation in studies associated with ESCC.\u003c/p\u003e\n\u003cp\u003eSubsequently, we screened core genes based on the interaction network, including KIF14, CENPF, CDK1, CDC6, CDC45, FOXM1, CKS1B, FEN1, HMMR, CDKN3, IGFBP3, CXCL1, COL4A2, COL4A1, ACVR1, COL3A1, COL1A2, COL5A2, FNI, LUM, CDH11, MMP2, MMP1. Using the TCGA-ESCC dataset as a basis, we analyzed the interrelationship of mRNA expressions among core genes (Figure 4A). These genes can be distinctively divided into two groups, namely, cell cycle and tumor EMT-related genes.\u003c/p\u003e\n\u003cp\u003eThe analysis of gene mRNA expression reveals that all selected genes are significantly up-regulated in tumor tissues compared to healthy esophageal tissue. Subsequently, we employed the Cox model, obtaining a prognostic model from the TCGA-ESCC dataset that can effectively predict patient outcomes. Although some studies have reported on gene-set-based prognostic models, their AUC values are not high, possibly due to their method of selectively using certain gene classes to build their prognostic models. Our model, while not incorporating some renowned genes associated with cancer prognosis like ARUKA and NEK2[32], nevertheless achieved the best result reported until now, with a 3-year AUC value of 0.98.\u003c/p\u003e\n\u003cp\u003eWe used the Cox algorithm to compute the gene set across multiple cancer databases. The results demonstrated that the gene set also has excellent prognostic value in ESCA, with a 5-year AUC value of 1, and its ACI value is just slightly higher than that of ESCC. This suggests that this gene set might be applicable to both ESCC and ESCA, indicating its high specificity across a wide range of cancers.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eCancer cells reorganize and reprogram host cells, and regulate the tumor-supporting environment by reshaping the vascular system and extracellular matrix. This dynamic process depends on the interactions between cancer cells and the resident or recruited non-cancerous cells in the tumor microenvironment[33]. Macrophages are highly plastic cells with various functions, including tissue development and balance, clearance of cell debris, elimination of pathogens, and regulation of inflammatory responses[34]. While the range of macrophage activation states is complex, it is generally simplified into two types: CLASSICAL M1-activated macrophages or ALTERNATIVE M2-activated macrophages[35]. M1 macrophage polarization is driven by exposure to GM-CSF, IFN-\u0026gamma;, TNF-\u0026alpha;, lipopolysaccharide (LPS), or other pathogen-associated molecular patterns[36]. It is crucial in the response to bacterial and viral pathogens and may possibly participate in anti-tumor immunity[37].\u0026nbsp;M2 macrophages play a crucial role in normal immune function and internal balance, such as stimulating Th2 responses, eradicating parasites, immune regulation, wound healing, and tissue regeneration. They promote tumor angiogenesis and metastasis by secreting factors such as VEGF and matrix metalloproteinases, which reshape the tumor microenvironment (TME), increase blood vessel formation, and promote tumor cell migration\u0026nbsp;[38, 39]. In fact, the number of M2 macrophages gradually increases in the early stages of esophageal cancer, and it can be clearly observed even at the hyperplasia stage[40, 41]. The long-term chronic inflammation leads to an imbalance in the immune microenvironment of esophageal cancer[42], which is consistent with what we observed in our network analysis (Fig.8).\u0026nbsp;Nearly half of the core genes show a significant positive correlation with the number of M2 cells, and the gene signature also has the most pronounced correlation with M2 macrophages. Recent studies also showed that M2 polarization of macrophages can enhance the expression of PD-L2 in tumor-associated macrophages, leading to immune evasion and tumor promotion through the PD-1 pathway. The blockade of the CCL2-CCR2 axis can significantly hinder the recruitment of tumor-associated macrophages, resulting in a positive response to treatment[43]. In our model, we observed a strong correlation between the core genes and PD1L2 expression (Fig.10A). In addition, we noticed another checkpoint gene, SIGLEC-15[44], which has strong tumor immune suppression ability, and it displays even stronger correlation with our core genes. This suggests that immunotherapy may yield better benefits in the treatment of high-risk patients. The prediction of ICB response based on TIDE supports our hypothesis (Fig.10B).\u003c/p\u003e\n\u003cp\u003eHowever, when we predicted the correlation between first-line chemotherapy and the prognostic model, we found that high-risk group patients were less sensitive to most traditional drugs (Fig. 9). But, the miraculous metformin, which has multiple abilities, showed high sensitivity in the high-risk group. This could be related to several factors, such as metformin\u0026rsquo;s ability to reduce tumor metabolism, improve the tumor microenvironment, and reactivate tumor immunity. It has been reported that metformin has the ability of reverse the polarization of M2 macrophages[45]. Studies have also shown that metformin can reduce the number of myeloid-derived suppressor cells and M2 macrophages in tumors[46]. Combined treatment of metformin with anti-PD-1/PD-L1 antibodies might bring more benefits to patients[47].\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This work was supported in part by grants from the Technology Commission Foundation of Shanxi Province (Grant No. 20210302124292, 20210302124296 and 20210302124091), Science and technology innovation plan of Shanxi Higher Education Institutions (Grant No. 2021L343, 2020L0373, 2021L342) and Changzhi Medical College doctoral research fund (Grant No. 521300).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability:\u003c/strong\u003e The\u0026nbsp;datasets used and/or analysed during the current study available from the corresponding author on reasonable request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAbnet CC, Arnold M, Wei WQ. Epidemiology of Esophageal Squamous Cell Carcinoma. Gastroenterology. 2018; 154: 360-73.\u003c/li\u003e\n\u003cli\u003eYang YM, Hong P, Xu WW, He QY, Li B. Advances in targeted therapy for esophageal cancer. Signal Transduct Target Ther. 2020; 5: 229.\u003c/li\u003e\n\u003cli\u003eMcCormack VA, Menya D, Munishi MO, Dzamalala C, Gasmelseed N, Leon Roux M, Assefa M, Osano O, Watts M, Mwasamwaja AO, Mmbaga BT, Murphy G, Abnet CC, et al. Informing etiologic research priorities for squamous cell esophageal cancer in Africa: A review of setting-specific exposures to known and putative risk factors. Int J Cancer. 2017; 140: 259-71.\u003c/li\u003e\n\u003cli\u003eHe S, Xu J, Liu X, Zhen Y. Advances and challenges in the treatment of esophageal cancer. Acta Pharm Sin B. 2021; 11: 3379-92.\u003c/li\u003e\n\u003cli\u003eZhang BH, Yang J, Jiang L, Lyu T, Kong LX, Tan YF, Li B, Zhu YF, Xi AY, Xu X, Yan LN, Yang JY. Development and validation of a 14-gene signature for prognosis prediction in hepatocellular carcinoma. Genomics. 2020; 112: 2763-71.\u003c/li\u003e\n\u003cli\u003eLu T, Xu R, Li Q, Zhao JY, Peng B, Zhang H, Guo JD, Zhang SQ, Li HW, Wang J, Zhang LY. Systematic profiling of ferroptosis gene signatures predicts prognostic factors in esophageal squamous cell carcinoma. Mol Ther Oncolytics. 2021; 21: 134-43.\u003c/li\u003e\n\u003cli\u003eYe J, Wu Y, Cai H, Sun L, Deng W, Liang R, Han A. Development and Validation of a Ferroptosis-Related Gene Signature and Nomogram for Predicting the Prognosis of Esophageal Squamous Cell Carcinoma. Front Genet. 2021; 12: 697524.\u003c/li\u003e\n\u003cli\u003eSong K, Gu B, Ma C, Wang B, Wang N, Yu R, Chen H. Epithelial-Mesenchymal Transition Gene Signature Is Associated with Neoadjuvant Chemoradiotherapy Resistance and Prognosis of Esophageal Squamous Cell Carcinoma. Dis Markers. 2022; 2022: 3534433.\u003c/li\u003e\n\u003cli\u003eShi X, Li Y, Pan S, Liu X, Ke Y, Guo W, Wang Y, Ruan Q, Zhang X, Ma H. Identification and validation of an autophagy-related gene signature for predicting prognosis in patients with esophageal squamous cell carcinoma. Sci Rep. 2022; 12: 1960.\u003c/li\u003e\n\u003cli\u003eBardou P, Mariette J, Escudie F, Djemiel C, Klopp C. jvenn: an interactive Venn diagram viewer. BMC Bioinformatics. 2014; 15: 293.\u003c/li\u003e\n\u003cli\u003eZhou Y, Zhou B, Pache L, Chang M, Khodabakhshi AH, Tanaseichuk O, Benner C, Chanda SK. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun. 2019; 10: 1523.\u003c/li\u003e\n\u003cli\u003eHochberg Y, Benjamini Y. More powerful procedures for multiple significance testing. Stat Med. 1990; 9: 811-8.\u003c/li\u003e\n\u003cli\u003eHan H, Shim H, Shin D, Shim JE, Ko Y, Shin J, Kim H, Cho A, Kim E, Lee T, Kim H, Kim K, Yang S, et al. TRRUST: a reference database of human transcriptional regulatory interactions. Sci Rep. 2015; 5: 11432.\u003c/li\u003e\n\u003cli\u003eBader GD, Hogue CW. An automated method for finding molecular complexes in large protein interaction networks. BMC Bioinformatics. 2003; 4: 2.\u003c/li\u003e\n\u003cli\u003eYi L, Wu G, Guo L, Zou X, Huang P. Comprehensive Analysis of the PD-L1 and Immune Infiltrates of m(6)A RNA Methylation Regulators in Head and Neck Squamous Cell Carcinoma. Mol Ther Nucleic Acids. 2020; 21: 299-314.\u003c/li\u003e\n\u003cli\u003eJin K, Qiu S, Jin D, Zhou X, Zheng X, Li J, Liao X, Yang L, Wei Q. Development of prognostic signature based on immune-related genes in muscle-invasive bladder cancer: bioinformatics analysis of TCGA database. Aging (Albany NY). 2021; 13: 1859-71.\u003c/li\u003e\n\u003cli\u003eXu F, Huang X, Li Y, Chen Y, Lin L. m(6)A-related lncRNAs are potential biomarkers for predicting prognoses and immune responses in patients with LUAD. Mol Ther Nucleic Acids. 2021; 24: 780-91.\u003c/li\u003e\n\u003cli\u003eLi T, Fu J, Zeng Z, Cohen D, Li J, Chen Q, Li B, Liu XS. TIMER2.0 for analysis of tumor-infiltrating immune cells. Nucleic Acids Res. 2020; 48: W509-W14.\u003c/li\u003e\n\u003cli\u003eXu L, Deng C, Pang B, Zhang X, Liu W, Liao G, Yuan H, Cheng P, Li F, Long Z, Yan M, Zhao T, Xiao Y, et al. TIP: A Web Server for Resolving Tumor Immunophenotype Profiling. Cancer Res. 2018; 78: 6575-80.\u003c/li\u003e\n\u003cli\u003eJiang Q, Sun J, Chen H, Ding C, Tang Z, Ruan Y, Liu F, Sun Y. Establishment of an Immune Cell Infiltration Score to Help Predict the Prognosis and Chemotherapy Responsiveness of Gastric Cancer Patients. Front Oncol. 2021; 11: 650673.\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, Liu J, Freeman GJ, Brown MA, et al. Signatures of T cell dysfunction and exclusion predict cancer immunotherapy response. Nat Med. 2018; 24: 1550-8.\u003c/li\u003e\n\u003cli\u003eHanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov. 2022; 12: 31-46.\u003c/li\u003e\n\u003cli\u003eZinatizadeh MR, Schock B, Chalbatani GM, Zarandi PK, Jalali SA, Miri SR. The Nuclear Factor Kappa B (NF-kB) signaling in cancer development and immune diseases. Genes Dis. 2021; 8: 287-97.\u003c/li\u003e\n\u003cli\u003eLi X, Ding F, Wang L, Chen H, Liu Z. Disruption of enhancer-driven S100A14 expression promotes esophageal carcinogenesis. Cancer Lett. 2022; 545: 215833.\u003c/li\u003e\n\u003cli\u003eZhu F, Willette-Brown J, Song NY, Lomada D, Song Y, Xue L, Gray Z, Zhao Z, Davis SR, Sun Z, Zhang P, Wu X, Zhan Q, et al. Autoreactive T Cells and Chronic Fungal Infection Drive Esophageal Carcinogenesis. Cell Host Microbe. 2017; 21: 478-93 e7.\u003c/li\u003e\n\u003cli\u003eWang Y, Zhu C, Wang Y, Sun J, Ling D, Wang L. Survival risk prediction model for ESCC based on relief feature selection and CNN. Comput Biol Med. 2022; 145: 105460.\u003c/li\u003e\n\u003cli\u003eNomburg J, Bullman S, Nasrollahzadeh D, Collisson EA, Abedi-Ardekani B, Akoko LO, Atkins JR, Buckle GC, Gopal S, Hu N, Kaimila B, Khoshnia M, Malekzadeh R, et al. An international report on bacterial communities in esophageal squamous cell carcinoma. Int J Cancer. 2022; 151: 1947-59.\u003c/li\u003e\n\u003cli\u003eWei J, Li R, Lu Y, Meng F, Xian B, Lai X, Lin X, Deng Y, Yang D, Zhang H, Li L, Ben X, Qiao G, et al. Salivary microbiota may predict the presence of esophageal squamous cell carcinoma. Genes Dis. 2022; 9: 1143-51.\u003c/li\u003e\n\u003cli\u003eHuang Y, Hong W, Wei X. The molecular mechanisms and therapeutic strategies of EMT in tumor progression and metastasis. J Hematol Oncol. 2022; 15: 129.\u003c/li\u003e\n\u003cli\u003eDeng JL, Xu YH, Wang G. Identification of Potential Crucial Genes and Key Pathways in Breast Cancer Using Bioinformatic Analysis. Front Genet. 2019; 10: 695.\u003c/li\u003e\n\u003cli\u003eMou PK, Yang EJ, Shi C, Ren G, Tao S, Shim JS. Aurora kinase A, a synthetic lethal target for precision cancer medicine. Exp Mol Med. 2021; 53: 835-47.\u003c/li\u003e\n\u003cli\u003eSu W, Hu H, Ding Q, Wang M, Zhu Y, Zhang Z, Geng Z, Lin S, Zhou P. NEK2 promotes the migration and proliferation of ESCC via stabilization of YAP1 by phosphorylation at Thr-143. Cell Commun Signal. 2022; 20: 87.\u003c/li\u003e\n\u003cli\u003eDeNardo DG, Ruffell B. Macrophages as regulators of tumour immunity and immunotherapy. Nat Rev Immunol. 2019; 19: 369-82.\u003c/li\u003e\n\u003cli\u003eVarol C, Mildner A, Jung S. Macrophages: development and tissue specialization. Annu Rev Immunol. 2015; 33: 643-75.\u003c/li\u003e\n\u003cli\u003eMosser DM, Edwards JP. Exploring the full spectrum of macrophage activation. Nat Rev Immunol. 2008; 8: 958-69.\u003c/li\u003e\n\u003cli\u003eJaguin M, Houlbert N, Fardel O, Lecureur V. Polarization profiles of human M-CSF-generated macrophages and comparison of M1-markers in classically activated macrophages from GM-CSF and M-CSF origin. Cell Immunol. 2013; 281: 51-61.\u003c/li\u003e\n\u003cli\u003eItaliani P, Boraschi D. From Monocytes to M1/M2 Macrophages: Phenotypical vs. Functional Differentiation. Front Immunol. 2014; 5: 514.\u003c/li\u003e\n\u003cli\u003eYang L, Zhang Y. Tumor-associated macrophages: from basic research to clinical application. J Hematol Oncol. 2017; 10: 58.\u003c/li\u003e\n\u003cli\u003eZhang X, Ji L, Li MO. Control of tumor-associated macrophage responses by nutrient acquisition and metabolism. Immunity. 2023; 56: 14-31.\u003c/li\u003e\n\u003cli\u003eJia Y, Zhang B, Zhang C, Kwong DL, Chang Z, Li S, Wang Z, Han H, Li J, Zhong Y, Sui X, Fu L, Guan X, et al. Single-Cell Transcriptomic Analysis of Primary and Metastatic Tumor Ecosystems in Esophageal Squamous Cell Carcinoma. Adv Sci (Weinh). 2023; 10: e2204565.\u003c/li\u003e\n\u003cli\u003eDinh HQ, Pan F, Wang G, Huang QF, Olingy CE, Wu ZY, Wang SH, Xu X, Xu XE, He JZ, Yang Q, Orsulic S, Haro M, et al. Integrated single-cell transcriptome analysis reveals heterogeneity of esophageal squamous cell carcinoma microenvironment. Nat Commun. 2021; 12: 7335.\u003c/li\u003e\n\u003cli\u003eYe H, Li X, Lin J, Yang P, Su M. CD98hc has a pivotal role in maintaining the immuno-barrier integrity of basal layer cells in esophageal epithelium. Cancer Cell Int. 2022; 22: 98.\u003c/li\u003e\n\u003cli\u003eYang H, Zhang Q, Xu M, Wang L, Chen X, Feng Y, Li Y, Zhang X, Cui W, Jia X. CCL2-CCR2 axis recruits tumor associated macrophages to induce immune evasion through PD-1 signaling in esophageal carcinogenesis. Mol Cancer. 2020; 19: 41.\u003c/li\u003e\n\u003cli\u003eWang J, Sun J, Liu LN, Flies DB, Nie X, Toki M, Zhang J, Song C, Zarr M, Zhou X, Han X, Archer KA, O\u0026apos;Neill T, et al. Siglec-15 as an immune suppressor and potential target for normalization cancer immunotherapy. Nat Med. 2019; 25: 656-66.\u003c/li\u003e\n\u003cli\u003ede Oliveira S, Houseright RA, Graves AL, Golenberg N, Korte BG, Miskolci V, Huttenlocher A. Metformin modulates innate immune-mediated inflammation and early progression of NAFLD-associated hepatocellular carcinoma in zebrafish. J Hepatol. 2019; 70: 710-21.\u003c/li\u003e\n\u003cli\u003eKang J, Lee D, Lee KJ, Yoon JE, Kwon JH, Seo Y, Kim J, Chang SY, Park J, Kang EA, Park SJ, Park JJ, Cheon JH, et al. Tumor-Suppressive Effect of Metformin via the Regulation of M2 Macrophages and Myeloid-Derived Suppressor Cells in the Tumor Microenvironment of Colorectal Cancer. Cancers (Basel). 2022; 14.\u003c/li\u003e\n\u003cli\u003eWei Z, Zhang X, Yong T, Bie N, Zhan G, Li X, Liang Q, Li J, Yu J, Huang G, Yan Y, Zhang Z, Zhang B, et al. Boosting anti-PD-1 therapy with metformin-loaded macrophage-derived microparticles. Nat Commun. 2021; 12: 440.\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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Esophageal squamous cell carcinoma, gene signature, prognosis, immunotherapy","lastPublishedDoi":"10.21203/rs.3.rs-3208103/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3208103/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eEsophageal squamous cell carcinoma (ESCC) is a highly aggressive upper gastrointestinal tumor with a 5-year survival rate of less than 20%. Therefore, developing new effective prognostic markers is of great clinical significance. In this study, we utilized datasets specific to ESCC and analyzed differentially expressed genes in each dataset. By conducting Venn analysis, we identified genes that exhibited significant differential expression across multiple datasets. Through gene interaction network analysis, we identified a core set of genes (23 genes) and established a prognostic model for ESCC using the COX algorithm (p=0.000245, 3-year AUC=0.98). The high-risk group of patients showed a significantly worse prognosis compared to the low-risk group. Furthermore, immune interaction network analysis revealed a strong association between increased risk values and an elevated presence of M2 macrophages within tumor tissues. Drug sensitivity analysis indicated that the high-risk group of patients exhibited poorer sensitivity to first-line chemotherapy drugs for ESCC. Notably, there was a significant positive correlation between the expression of core genes and immune checkpoint genes such as SIGLEC15, PDCD1LG2, and HVCR2. The high-risk group exhibits decreased Tumor Immune Dysfunction and Exclusion (TIDE) values, indicating that immune checkpoint blockade therapy might result in more favorable outcomes for these individuals. The immune checkpoint blockade (ICB) therapy may potentially yield better outcomes for these patients. In summary, through comprehensive bioinformatics analysis, we have established a highly effective prognostic model consisting of 23 genes for ESCC. An increased risk score in this model indicates a stronger infiltration of M2 macrophages and poorer sensitivity to chemotherapy drugs. Moreover, immune checkpoint blockade therapy may hold greater benefits for patients in the high-risk group.\u003c/p\u003e","manuscriptTitle":"Development of a Novel Prognostic Model for Esophageal Squamous Cell Carcinoma: Insights into Immune Cell Interactions and Drug Sensitivity","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2023-08-04 23:36:29","doi":"10.21203/rs.3.rs-3208103/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c1c01217-5161-4ce2-8965-511cb5049a52","owner":[],"postedDate":"August 4th, 2023","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":23688925,"name":"Biological sciences/Cancer"},{"id":23688926,"name":"Biological sciences/Cancer/Cancer genomics"},{"id":23688927,"name":"Biological sciences/Cancer/Cancer therapy"},{"id":23688928,"name":"Biological sciences/Cancer/Gastrointestinal cancer"},{"id":23688929,"name":"Biological sciences/Cancer/Tumour biomarkers"}],"tags":[],"updatedAt":"2024-04-15T04:13:19+00:00","versionOfRecord":{"articleIdentity":"rs-3208103","link":"https://doi.org/10.1080/07357907.2024.2340576","journal":{"identity":"cancer-investigation","isVorOnly":true,"title":"Cancer Investigation"},"publishedOn":"2024-04-14 04:13:19","publishedOnDateReadable":"April 14th, 2024"},"versionCreatedAt":"2023-08-04 23:36:29","video":"","vorDoi":"10.1080/07357907.2024.2340576","vorDoiUrl":"https://doi.org/10.1080/07357907.2024.2340576","workflowStages":[]},"version":"v1","identity":"rs-3208103","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3208103","identity":"rs-3208103","version":["v1"]},"buildId":"_2-kVJe1T_tPrBINL-cwx","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}