Subtyping The Epstein-Barr Virus And Generating A Prognostic Model For Nasopharyngeal Carcinoma Based On Transcriptome Sequencing Analysis

Background:Epstein-Barr virus(EBV)widely infects human beings,causing a variety of diseases including malignant tumors,especially are closely related to nasopharyngeal carcinoma(NPC).EBV nucleic acids can be detected in 98%NPC tumors.Nowadays,many EBV genomes have been sequenced.By comparing the genome sequences of EBVs obtained from different tumor types,the association between EBV variants and specific tumor types started to emerge.Variations in EBV genome may play a significant role in its tumorigenic process,thereby affecting the biological behavior of specific types of tumors.However,knowledge about NPC associated EBV variants are still limited.NPC is a highly invasive and metastatic cancer,with diverse molecular characteristics and clinical outcomes.The tumor-node-metastasis(TNM)staging system is currently a guidance for treatment and is a fundamental determinant of prognostic predictions.However,the value of this system for assessing the prognosis of NPC is limited,and it may no longer fully meet the clinical needs.Objective:The research work of this Ph.D.project was based on transcriptome sequencing in NPC,and had two purpose.1.To discover and identify EBV variations related to NPC,and to analyze the relationship between EBV subtypes and clinical characteristics,prognosis of NPC patients.This study initially revealed the biological function of EBV variations.2.To screen and establish a novel gene signature that could predict prognosis of patients with NPC.Methods:This project included two groups of NPC patients.The first group of 67 cases,mainly involved in transcriptome sequencing and related analysis;The second group of 32 cases,mainly participated in the verification of EBV variation.The two groups of NPC cases were independent.Transcriptome sequencing were performed for each tumor and adjacent normal sample of the first group.EBV variants were analyzed from this sequencing data.The identified EBV variations were further validated by applying BaseScope and immunohistochemistry staining in the second group of 32 formalin fixed paraffin embedded(FFPE)NPC tissues.Compared to 85 published EBVs isolated from other diseases by cluster analysis,we analyzed the EBV subtypes associated with NPC.We also analyzed the relationship between different subtypes and clinical characteristics,prognosis of NPC patients.In addition,we established NPC cell lines stably expressing the above EBV variation genes and explored the biological function of it using conventional cell molecular biology method.In the study of prognostic model construction,a total of 60 NPC patients in the first group of 67 patients participating in transcriptome sequencing and with complete follow-up information were enrolled.They were divided into recurrent/metastatic(RM)group and no recurrent/metastatic(no-RM)group based on the follow-up results.Machine learning and Cox proportional hazards regression model were used to identify gene signature for NPC prognosis.GSEA and spearman correlation analysis were used to evaluate the function of gene signature.Results:1.Transcriptome sequencing suggested that variations existed in the genome of EBV,among which the EBER2,LMP2 had the highest SNP density.The variation of these genes was detected in 65/66 and 64/66 NPC tumor tissues,respectively.2.BaseScope and immunohistochemical co-staining techniques verified the EBER2 variants discovered in this study and showed that the EBVs were specifically found in tumor(epithelial)cells by visualizing the variations at the single-cell level.3.The EBV isolates in our study and 85 published EBV isolates were clustered according to the EBER2 variation sites.EBV could be divided into two major subtypes:EB-1 and EB-2.Furthermore,EB-2 subtype could be further classified according to the EBER2 7123 genotype.Notably,survival analysis found that the three subtypes had different survival outcomes,among which EB-2-A>T subtypes had the worst prognosis(Log-rank,p=0.026).Moreover,results showed that EB-1 subtype were significantly enriched in NPC(chi-square test,p<0.0001),while EBER2 EB-2 has higher frequency in NPC non-endemic areas and were related to T stage(chi-square test,p<0.05).4.Analyzing mRNA expression profile in host,we found that EB-1 and EB-2 subtype involved in different biological processes,namely "Epithelial mesenchymal transition" and "MYC targets v1",etc.,suggesting that they were involved in NPC carcinogenesis through different biological processes.5.Another gene with high variation frequency is LMP2B.In vitro analysis showed NPC specific subtype LMP2B-C666 could promote NPC cells proliferation compared with LMP2B-Raji,a EBV subtype representing lymphoma.The study of constructing prognostic model of NPC mainly focused on host gene expression.6.A total of 13 significant genes between RM group and no-RM group were identified by machine learning,including CES4A?GIMAP1.GIMAP5?GLB1L?LGR5?LOC730098?MLF1?MTHFD1L?MYLPF?NUP160?SIRPB1?TCN2?TMEM265?U2AF1L5.7.Based on these genes,a Cox proportional hazards regression model was used to identify a 4-mRNA signature(U2AF1L5,TMEM265,GLB1L and MLF1)model.According to the model,the risk score of each patient was calculated,and the patients were divided into high-risk group and low-risk group.8.Receiver operating characteristic(ROC)and Kaplan-Meier(K-M)analysis indicated this model had good prognostic value for NPC.The overall survival(OS)and progression-free survival(PFS)of patients in high-risk group were significantly shorter than that of patients in low-risk group(p=0.00126,p=0.000059 respectively).The AUC of 4-mRNA signature was higher than that of T stage and N stage(OS:0.893 vs 0.619,0.582,PFS:0.86 vs 0.538,0.622).9.The risk score was an independent predictor for OS and PFS in multivariate Cox regression analyses(OS:HR=14.501,p=0.012,PFS:HR=13.752,p<0.001).10.The 4-mRNA signature was closely associated with proliferation and immune response.Conclusion:1.EBV genome variations exited in NPC.The EBER2 and LMP2 genes had the highest SNP density in NPC patients.In addition,we conclusively showed that EBV variations was present in tumor cells instead of infiltrating lymphocytes.According to EBER2 variation,EBV was divided into two subtypes:EB-1 and EB-2.The subtypes were related to clinical characteristics and prognosis of NPC patients,and may have different interactions with the host.2.A novel 4-mRNA signature which could predict prognosis of NPC patients was established by analyzing transcriptome data.