Application of GALNT12 in preparation of esophageal squamous cell carcinoma diagnosis and treatment product

Abstract

The invention discloses application of GALNT12 in preparation of esophageal squamous cell carcinoma diagnosis and treatment products. The invention discloses the significant overexpression of the GALNT12 gene in ESCC, and the strong correlation between the high expression level of the GALNT12 and the poor prognosis of a malignant tumor patient. Furthermore, through comprehensive single gene set enrichment analysis (GSEA), key effects of GALNT12 in a number of biologically important pathways, including ascorbic acid and alternative metabolism, chemical oncogen-related DNA modifications, drug metabolism through cytochrome P450, extracellular matrix (ECM) receptor interactions, and IL-17 signaling pathways, have been determined.

Description

Technical Field

[0001] The present application relates to the field of biomedical technology, and specifically to the use of a protein in preparing a diagnostic and therapeutic product for esophageal squamous cell carcinoma. Background Art

[0002] The International Agency for Research on Cancer (IARC) recently revised its global cancer statistics, revealing that esophageal cancer (ESCA) is the seventh most common cancer worldwide, with approximately 604,000 new cases, and the sixth leading cause of cancer-related deaths, with an estimated 544,000 deaths. ESCA can be divided into two major histological types: esophageal squamous cell carcinoma (ESCC) and esophageal adenocarcinoma (EA), of which ESCC accounts for 85% of all cases. Despite advances in various treatment modalities, including surgical resection, endoscopic surgery, minimally invasive surgery, chemotherapy, radiotherapy, immunotherapy, and targeted therapy, which have collectively improved the survival outcomes of ESCC patients, the 5-year overall survival (OS) rate remains less than 20%. Many studies have found that several prognostic factors, such as tumor stage, smoking, and alcohol consumption, can be used to predict the OS of ESCC patients. Despite extensive research to understand the etiology and progression of ESCC, its underlying causes remain largely unknown. Discovering new prognostic biomarkers and therapeutic targets could provide valuable insights into the molecular mechanisms driving the initiation and progression of ESCC.

[0003] Advances in genomic technologies, including microarrays and high-throughput sequencing, have facilitated the frequent identification of genes and their co-expression patterns that contribute to ESCC progression. The field of integrated bioinformatics has emerged as a valuable tool for biomarker discovery from genomic data sources, such as the Gene Expression Omnibus Database (GEO). In many studies, a large number of gene chips have been used to identify key molecular determinants of ESCC. For example, Yang et al. used an integrated bioinformatics strategy to reveal several key genes and potential therapeutic drugs associated with ESCC. The KIF4A gene was identified as a prognostic biomarker for ESCC. Another group of genes, including SPP1, MMP12, COL10A1, and COL5A2, were confirmed to play a role in ESCC progression by integrated bioinformatics analysis. In addition, the CDC6 gene has also been considered a new candidate gene for ESCC diagnosis and therapeutic intervention. Although these studies have elucidated some of the molecular basis of ESCC, the clarity of the interpretation of the disease mechanism is still hampered by the lack of comprehensive genetic evaluation and extensive patient samples. Summary of the invention

[0004] This application clarifies the use of GALNT12 in the preparation of esophageal squamous cell carcinoma diagnosis and treatment products.

[0005] The inventor's research method is mainly: differentially expressed genes (DEGs) were identified using three data sets (GSE53625, GSE67269 and GSE23400-GPL96). At the same time, a gene co-expression network was constructed based on weighted gene co-expression network analysis (WGCNA) of the GSE23400-GLP97 data set. The machine learning algorithm further identified the most critical genes. In addition, we used GSE161533 and GSE38129 data sets to verify the expression and diagnostic potential of the hub genes. Survival analysis and gene set enrichment analysis (GSEA) revealed the prognostic value and potential function of the hub genes, respectively.

[0006] The main results of the study were: 240 differentially expressed genes were identified (103 upregulated and 137 downregulated). At the same time, WGCNA identified 209 genes associated with ESCC. Subsequently, the machine learning algorithm identified four central genes including KIF14, GALNT12, MGLL and EMP1. In addition, their expression differences and potential as diagnostic biomarkers for ESCC were verified. Survival analysis showed that increased expression levels of GALNT12 were associated with poor prognosis in ESCC patients. GSEA depicted the involvement of GALNT12 in key biological pathways. Thus, GALNT12 was identified as a new potential diagnostic and prognostic marker for ESCC.

[0007] Therefore, this application provides the following practical technical solutions:

[0008] First, the use of GALNT12 in the preparation of products for the diagnosis / screening of esophageal squamous cell carcinoma, wherein GALNT12 is highly expressed in patients with esophageal squamous cell carcinoma, and higher GALNT12 expression levels are correlated with poor prognosis in patients with esophageal squamous cell carcinoma.

[0009] Optionally, the product for diagnosis / screening of esophageal squamous cell carcinoma is a kit, a chip, a test paper or a computer program product.

[0010] In a second aspect, GALNT12 is used to prepare a product for treating esophageal squamous cell carcinoma, wherein esophageal squamous cell carcinoma is inhibited by reducing the expression of the GALNT12 gene.

[0011] Optionally, the product for treating esophageal squamous cell carcinoma includes protein kinase inhibitors, RNAi and CRISPR systems.

[0012] The third aspect is the use of GALNT12 in the preparation of ascorbic acid and alternative metabolic drugs. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1This is a schematic diagram of the workflow of the research method for this application;

[0014] Figure 2 Identification diagram of differentially expressed genes (DEGs) between ESCC and normal tissues, A. Volcano plot of DEGs of GSE53625. B. Volcano plot of DEGs of GSE67269. C. Volcano plot of DEGs of GSE23400-GLP96. D. Venn diagram of up-regulated DEGs based on three GEO datasets. E. Venn diagram of down-regulated DEGs based on three GEO datasets;

[0015] Figure 3 The following are the gene co-expression network analysis diagrams: A. Sample dendrogram and feature heat map of the dataset GSE23400-GLP97. B. Scale-free fit index and average connectivity of various soft threshold capabilities (β). C. Cluster dendrogram of genes with variance greater than 0.25. D. Heat map of module-trait relationship. E. Scatter plot of gene module membership and gene significance correlation in turquoise modules. F. Venn diagram of ESCC key module genes and DEGs.

[0016] Figure 4 Screening of core gene graph for machine learning, AB. LASSO regression algorithm was used to identify core genes. CD. RF identified core genes with importance scores greater than 5. E. 12 core genes were identified by SVM-RFE analysis. F: Venn diagram showing the intersection of genes obtained by the three algorithms;

[0017] Figure 5 Figure 1 shows the expression level and diagnostic value of core genes. A and B show the difference in expression of four core genes in the ESCC and normal group training sets (GSE161533 and GSE38129). C shows the ROC curves of the four core genes in the ESCC group and the normal group.

[0018] Figure 6 Figure 1 shows the prognostic core genes and the identification of GSEA. A. Kaplan-Meier overall survival (OS) curve comparison of the high-expression group and the low-expression group of the four core genes (MGLL, GALNT12, KIF14 and EMP1) in ESCC patients. B. GSEA analysis of the GALNT12 gene. DETAILED DESCRIPTION

[0019] Esophageal squamous cell carcinoma (ESCC) is a highly lethal cancer whose molecular basis remains unclear. In this study, bioinformatics methods were used to reveal key genes and pathways associated with ESCC and identify prognostic markers. The inventors' research group selected six gene expression profile datasets from GEO, ensuring that each dataset contained more than 40 samples. First, differentially expressed genes (DEGs) between ESCC and normal esophageal tissue were identified in three training GEO datasets, each with a sample size of more than 100. At the same time, weighted gene co-expression network analysis (WGCNA) was used to identify key modules and central genes in the pathogenesis of ESCC based on another GEO dataset. To further narrow the list of candidate genes, the inventors used three different machine learning algorithms.

[0020] To validate our findings, we analyzed the expression levels of hub genes and their diagnostic value using two additional GEO datasets. Furthermore, we delved into the relationship between the expression of these hub genes and the OS rate of ESCC patients. Finally, we predicted the potential pathways involved in the function of hub genes. Our comprehensive study not only lays the foundation for future exploration of the molecular basis of ESCC tumorigenesis, but also provides clear insights into potential biomarkers and therapeutic targets that may play a key role in the clinical management of this condition.

[0021] The specific experimental process is introduced below:

[0022] 1. Materials and Methods

[0023] 1.1 Data download and preprocessing

[0024] To obtain specific ESCC mRNA expression data, we conducted a targeted search of the GEO database using the keyword search: ‘esophageal squamous cell carcinoma’, ‘Homo sapiens’, and ‘Expression profiling by array’. The systematic search yielded five relevant GSE datasets, each with more than 40 samples. The detailed information of the six GEO datasets are as follows: GSE53625 (GPL18109; 179 ESCC samples and 179 normal samples), GSE67269 (GLP571; 73 ESCC samples and 73 normal samples), GSE23400 (GPL96; 53 ESCC samples and 53 normal samples, GPL97; 51 ESCC samples and 51 normal samples), GSE38129 (GLP571; 30 ESCC samples and 30 normal samples) and GSE161533 (GLP570; 28 ESCC samples and 28 normal samples). In preprocessing, we used the platform annotation file to convert the probe expression matrix into a gene expression matrix (R software version: 3.4.6). Probes that could not be accurately matched to specific gene symbols were excluded from the analysis. When multiple probes corresponded to the same gene, the average of these probes was used as the expression level of that specific gene. The workflow is as follows Figure 1 shown.

[0025] 1.2 Analysis of gene differential expression

[0026] To identify differentially expressed genes (DEGs) between ESCC and normal groups, we analyzed the gene expression matrix files (GSE53625, GSE67269, and GSE23400-GPL96) using the “Limma” package in R software (version: 3.4.6). To meet the requirements of DEGs, genes were screened based on |log2FC| greater than 1 and adjusted p-value less than 0.05. Subsequently, we used the “ggplot2” package to generate volcano plots. Afterwards, we compared and synthesized the intersection of our differential analysis results using the Venn tool (https: / / jvenn.toulouse.inrae.fr / app / example.html).

[0027] 1.3WGCNA

[0028] To explore the relationship between genes and phenotypes, we constructed a gene co-expression network based on the GSE23400-GLP97 dataset using the “WGCNA” package in R software. Initially, we evaluated the variance of the expression level of each gene and excluded genes with variances below 0.25 and focused on genes with significant expression patterns. Subsequently, we clustered all samples using the flashClust tool in R to detect and exclude potential outliers. We then calculated the Pearson correlation coefficients for all pairwise gene combinations to assess the similarity of their expression levels and obtain the correlation matrix. To convert the correlation matrix into a weighted neighborhood matrix, we applied a soft threshold function and selected an optimal soft threshold that maximized the consistency of gene correlations and had scale-free properties. After determining the optimal soft threshold (β), we calculated the adjacency matrix and then converted it into a topological overlap matrix (TOM). We set a minimum cluster size of 100 genes and used the TOM-based differential average linkage hierarchical clustering method to divide these gene clusters. Different gene modules are represented by branches of different colors in the clustering tree. To further explore the relationship between these gene modules and clinical features, we calculated the correlation between them. In addition, we calculated gene significance (GS) and gene module identity (MM) to evaluate the importance of genes in their corresponding modules. Genes with |GS|>0.50 and |MM|>0.80 were considered candidate key genes. Finally, the candidate key genes were intersected with DEGs using the online "Venn" tool.

[0029] 1.4 Machine Learning

[0030] We used three machine learning algorithms (least absolute shrinkage and selection operator (LASSO) regression, random forest (RF), and support vector machine recursive feature elimination (SVM-RFE)) to screen the ESCC core genes in the intersection of the above DEGs and WCGNA. We used the “glmnet” package for lasso regression analysis. The “randomForest” and “arSelRF” packages in R were used for RF analysis. We used the “e1071”, “kernlab”, and “caret” packages for SVM-RFE analysis. The intersection genes of LASSO, RF, and SVM-RFE were considered as candidate core genes in ESCC.

[0031] 1.5 Validation of core gene expression levels and diagnostic potential

[0032] To verify the expression levels and diagnostic significance of the core genes screened in ESCC and normal tissues, we used the GSE161533 and GSE38129 datasets in GEO as validation sets. To visually represent gene expression levels, box plots were generated using the online Sangerbox software (http: / / vip.sangerbox.com / ). In addition, to evaluate the diagnostic value of the core genes, ROC curves were constructed.

[0033] 1.6 Survival analysis of core genes

[0034] To investigate the relationship between the expression levels of the core genes and the survival of ESCC patients, as well as their potential as therapeutic targets and prognostic indicators, we performed Kaplan-Meier survival curve analysis using the R package “surviving”.

[0035] 1.7 Single gene set enrichment analysis (GSEA)

[0036] To gain insight into the biological functions of the core genes, we performed single gene set enrichment analysis and visualization of the results using the R packages “clusterProfiler”, “enrichplot”, and “stringr”. Specifically, all samples were divided into two groups, namely, low and high expression groups of the core genes. Subsequently, we used GSEA to identify significant differences in KEGG pathways between the low and high expression groups. In addition, we performed Pearson correlation analysis to assess the relationship between the core genes and each of the other genes in our combined dataset. The genes were then prioritized based on their correlation scores. These priority genes served as the target gene set for further analysis, and the target gene set included the KEGG signaling pathway. Finally, we used a p-value < 0.05 as the statistical significance threshold to determine which KEGG signaling pathways were significantly enriched in our target gene set.

[0037] 2. Results

[0038] 2.1 Identification of DEGs

[0039] We identified 610 samples in three datasets (GSE53625, GSE67269, and GSE23400-GLP96), including 305 ESCC samples and 305 normal samples. Volcano plot ( Figure 2 AC) shows all DEGs of the three datasets. Based on the screening criteria of |log2 FC|>1 and adjusted p-value<0.05, 2,195 DEGs were successfully identified, of which 881 up-regulated and 1,314 down-regulated genes were obtained by analyzing the GSE53625 dataset ( Figure 2A). For the GSE67269 dataset, 865 DEGs were observed, including 441 up-regulated and 424 down-regulated genes ( Figure 2 B). For the GSE23400-GLP96 dataset, 390 DEGs were identified, including 185 up-regulated and 205 down-regulated genes ( Figure 2 C). Venn analysis was used to screen the DEGs common to the three datasets. Among them, 240 DEGs were identified from the three datasets, including 103 upregulated genes ( Figure 2 D) and 137 down-regulated genes ( Figure 2 E).

[0040] 2.2 Network construction and identification of key modules

[0041] To find the gene modules most associated with ESCC features, we constructed a gene co-expression network using WGCNA. By clustering samples to check for outliers, all samples in the GSE23400-GLP97 dataset met the criteria and did not need to be removed ( Figure 3 A). To ensure that the connections between genes follow a scale-free network distribution, we calculated the scale-free fit index and average connectivity. When R2 = 0.90, the soft threshold was 5, ensuring that the network was close to the scale-free topology ( Figure 3 B). WGCNA constructed a total of 4 co-expression modules ( Figure 3 CE). We plotted a heat map of the module-clinical feature relationship to evaluate the relevance of all modules to ESCC and found that the turquoise module (r = 0.86, p = 4.9e-29) had the highest positive correlation with ESCC ( Figure 3 E), so we selected this module for further analysis. The intersection of ESCC key module genes and DEGs was drawn through the Venn diagram, and 25 intersection genes were obtained ( Figure 3 F).

[0042] 2.3 Identification of core genes

[0043] Then, we used three machine learning algorithms (LASSO, RF, and SVM-RFE) to further screen the core genes from the 25 intersection genes. The LASSO algorithm identified five core genes, including DTL, EMP1, GALNT12, KIF14, and MGLL ( Figure 4 AB). RF identified nine core genes with importance scores greater than 5 ( Figure 4 CD). SVM identified twelve core genes ( Figure 4 E). Finally, we obtained the intersection results of the three machine learning methods through the Venn diagram and obtained four core genes (KIF14, GALNT12, MGLL and EMP1) ( Figure 4 F).

[0044] 2.4 Validation and diagnosis of core gene value

[0045] In the training sets (GSE161533 and GSE38129), three core genes (MGLL, GALNT12, and EMP1) were significantly overexpressed in ESCC compared with the normal group, while KIF14 was significantly underexpressed in the ESCC group (p<0.001, Figure 5 AB). The ROC curve showed that the areas under the curve (AUC) of the four core genes (MGLL, GALNT12, KIF14, and EMP1) were 0.993, 0.990, 0.989, and 0.986, respectively ( Figure 5 C).

[0046] 2.5 Identification of prognostic core genes and GSEA

[0047] To further identify key prognostic genes, overall survival (OS) analysis was performed based on GSE53625. Among the four core genes, high expression of GALNT12 was associated with poor prognosis in ESCC patients (p = 0.011, Figure 6 A). These results suggest that GALNT12 may be considered as an oncogene and prognostic gene in glioma. In addition, we performed single gene enrichment analysis (GSEA) to more accurately understand the function of the core gene GALNT12. Based on the specified screening criteria, we identified the top ten KEGG pathways in the enrichment analysis. For GALNT12, the enriched pathways included ascorbate and alternative metabolism, chemical carcinogenesis-DNA adducts, drug metabolism-cytochrome P450, extracellular matrix-receptor interactions, and IL-17 signaling pathways ( Figure 6 B).

[0048] 3. Conclusion Analysis

[0049] In the past few years, a large number of studies have been conducted by those skilled in the art to try to unravel the pathological mechanisms of esophageal squamous cell carcinoma (ESCC). Despite these efforts, the incidence and mortality of the disease continue to rise worldwide. The study of the inventor's research group is different from previous studies, which are usually limited to a single cohort or a small sample size. We adopted a broader strategy: analyzing six robust gene expression datasets, each with a sample size of more than 20 pairs from different research groups, to comprehensively investigate the key genes and biological pathways associated with ESCC.

[0050] In this study, we first identified 240 DGEs (103 upregulated and 137 downregulated) based on three GEO datasets. Meanwhile, we identified 209 genes significantly associated with ESCC using WCGNA. Subsequent cross-analysis with our identified DEGs yielded 25 intersection genes. Based on this, we identified the most critical genes, including KIF14, GALNT12, MGLL, and EMP1, using three machine learning algorithms. We then evaluated their expression in two additional GEO datasets and found significant differences between ESCC and controls, highlighting their diagnostic potential for ESCC. Our survival analysis revealed that GALNT12 overexpression was significantly associated with a worse prognosis in ESCC patients. In addition, GSEA highlighted the involvement of GALNT12 in several key pathways, such as ascorbate and alternative metabolism, DNA alterations caused by chemical carcinogens, cytochrome P450-mediated drug metabolism, extracellular matrix (ECM) receptor interactions, and IL-17 signaling pathways. These findings provide key insights into the complex molecular mechanisms of ESCC and its progression.

[0051] Since the advent of high-throughput sequencing and microarray technologies, bioinformatics methods have become a powerful tool for identifying hub genes. Many studies have used GEO datasets to reveal biomarkers associated with poor prognosis and survival outcomes in ESCC. For example, by integrating gene expression microarrays (GSE20347, GSE38129) and gene methylation microarrays (GSE52826), six aberrantly methylated-differentially expressed genes (KIF14, CDK1, AURKA, LCN2, TGM1, and DSG1) and their related pathways in ESCC were identified. In another study, three microarray datasets (GSE20347, GSE38129, and GSE67269) were analyzed to reveal six hub genes that were highly expressed in ESCC tissues. Wang et al. identified ten hub genes based on GSE26886, four of which were associated with overall survival in ESCC and EA patients. In addition, Shi et al. analyzed RNA microarray data (GSE53625) and developed a novel autophagy-related gene signature that enhanced the prediction of clinical outcomes in ESCC patients. Large-scale transcriptional profiling studies, including GSE17351, GSE20347, GSE23400, GSE38129, GSE77861, GSE161533, and GSE100942, identified nine hub genes (CORO1A, CD180, SASH3, CD52, CD300A, CD14, DUSP1, KIF14, and MCM2) that play key roles in key biological pathways for ESCC progression. Recently, a novel prognostic biomarker (EXOC3L1) was identified and validated using multiple datasets (UCSC Xena, GSE53625, and GSE161533), demonstrating its correlation with immune infiltration in ESCC. In our study, we identified and validated four genes (KIF14, GALNT12, MGLL, and EMP1) implicated in ESCC based on six GEO datasets (GSE53625, GSE23400, GSE67269, GSE161533, and GSE38129), further deepening our understanding of the genetic landscape of this malignancy. Compared with single dataset studies and small sample sizes, this study may provide more accurate results through integrated bioinformatics analysis.

[0052] Kinesin family member 14 (KIF14) is a member of the kinesin superfamily and plays a key role in chromosome segregation and cytoplasmic division. Multiple studies have implicated KIF14 in the development and progression of various cancers. Specifically, KIF14 has been shown to promote tumor growth and metastasis as an independent predictor of poor prognosis in human gastric cancer. Recent studies have revealed the tumor suppressor effect of exosome-derived miR-154-5p, which targets KIF14 and thus inhibits the progression and angiogenesis of ESCC. In addition, LETM1 silencing reduces the proliferation, invasion, migration, and angiogenesis of ESCC via KIF14. However, our study found that the expression of the KIF14 gene was downregulated in ESCC, a finding that requires further investigation to elucidate its clinical and biological significance.

[0053] Monoacylglycerol lipase (MGLL) is a key enzyme in lipid metabolism, responsible for the conversion of triglycerides to free fatty acids. This enzyme has attracted attention due to its association with cancer progression, especially in the context of tumor invasion and metastasis. The importance of MGLL in oncology is further emphasized by studies that have shown that its knockdown can effectively reduce cell proliferation and metastatic potential in lung adenocarcinoma. In addition, pharmacological inhibition of MGLL by ABX-1431 has been shown to impede the progression of endometrial adenocarcinoma and overcome progesterone resistance, highlighting its therapeutic potential. Interestingly, MGLL also acts as a regulatory switch for CB2 / TLR4-dependent macrophage activation, suggesting its multifaceted role in the tumor microenvironment and its promise as a target for cancer therapy. This study revealed for the first time that the MGLL gene is aberrantly overexpressed in ESCC and has potential diagnostic value. However, there was no significant correlation between MGLL expression and the prognosis of ESCC patients. Further investigation is needed to fully elucidate the role of MGLL in the pathogenesis of ESCC and its clinical significance.

[0054] Epithelial membrane protein 1 (EMP1) is a member of the epithelial membrane protein family and plays a key role in cellular functions such as adhesion, proliferation, apoptosis, differentiation, tumorigenesis, and metastasis. For example, overexpression of EMP1 in ovarian cancer has been associated with enhanced tumor cell proliferation, invasion, and epithelial-mesenchymal transition (EMT) through the MAPK signaling pathway. Conversely, deficiency of EMP1 in bladder cancer cells has been shown to increase migration and induce resistance to ferroptosis and oxidative stress, thereby promoting metastasis through PPARG. In addition, SMYD3 stimulates gastric cancer cell proliferation by reducing EMP1 expression in an H4K20me3-dependent manner. A meta-analysis combined with graph clustering techniques identified EMP1 as a potential prognostic marker for ESCC based on the analysis of datasets GSE20347 and GSE23400. In addition, previous studies have shown that the expression of EMP1 is downregulated in ESCC compared with normal esophageal tissue. In contrast, our study presents novel findings that EMP1 is significantly overexpressed in ESCC and is not associated with the prognosis of ESCC patients, which warrants further investigation to understand the impact of EMP1 in disease progression and patient outcomes.

[0055] Polypeptide N-acetylgalactosaminyltransferase 12 (GALNT12) is a member of a broad family of hexosyltransferases that plays a key role in the initial stages of mucin-type O-glycosylation. Perturbations in the enzymatic activity of GALNT12 may lead to dysregulated glycosylation, a phenomenon associated with a range of cellular processes including growth, differentiation, transformation, adhesion, and metastasis, all of which are critical in tumorigenesis. Specifically, GALNT12 has been shown to inhibit bone-specific metastasis of prostate cancer by regulating O-glycosylation of BMPR1A, thereby activating the BMP signaling pathway. Furthermore, GALNT12 enhances the invasiveness of glioblastoma in vitro by activating the Akt signaling cascade. Our findings reveal for the first time a significant overexpression of the GALNT12 gene in ESCC and a strong correlation between high GALNT12 expression levels and a poor prognosis in patients with this malignancy. In addition, through comprehensive single gene set enrichment analysis (GSEA), we have identified the key roles of GALNT12 in multiple biologically important pathways, including ascorbate and alternative metabolism, chemical carcinogen-related DNA modifications, drug metabolism through cytochrome P450, extracellular matrix (ECM) receptor interactions, and IL-17 signaling pathways. These findings and conclusions also provide valuable information for further elucidating the complex role of GALNT12 in the pathogenesis of ESCC and potential therapeutic targets.

Claims

1. Use of GALNT12 in the preparation of a product for diagnosis / screening of esophageal squamous cell carcinoma, wherein: GALNT12 is highly expressed in patients with esophageal squamous cell carcinoma, and higher GALNT12 expression levels are correlated with poor prognosis in patients with esophageal squamous cell carcinoma.

2. The use according to claim 1, characterized in that The product for diagnosis / screening of esophageal squamous cell carcinoma is a kit, a chip, a test paper or a computer program product.

3. Use of GALNT12 in the preparation of a product for treating esophageal squamous cell carcinoma, wherein: Inhibition of esophageal squamous cell carcinoma by reducing GALNT12 gene expression.

4. The use according to claim 3, characterized in that The products for treating esophageal squamous cell carcinoma include protein kinase inhibitors, RNAi and CRISPR systems.

5. Use of GALNT12 in the preparation of ascorbic acid and alternative metabolic drugs.

Images