Identification Of Biodiagnostic Markers Specific To The Tumor Microenvironment Based On The Cell Atlas

The occurrence and development of tumors are closely related to the microenvironment of tumor cells.The tumor microenvironment not only includes the structure,function and metabolism of the tumor tissue,but also has a close relationship with the internal environment of tumor cells.In recent years,due to the research progress of tumor cytology and molecular biology,people have a deeper understanding of the relationship between tumors and their microenvironment.This is not only important for understanding the occurrence,development and metastasis of tumors,but also plays an important role in the early diagnosis,prevention and prognosis of tumors.Objective: The tumor microenvironment is highly heterogeneous among patients,and only a small proportion of patients respond to a specific immunotherapy.Therefore,there is a need to better understand the immune microenvironmental landscape and its characteristics in tumor patients,and to identify microenvironment-related subtypes and immune molecular signatures.To identify tumor microenvironment-specific biomarkers based on cell atlases,in order to provide a new entry point for diagnosis and treatment in clinical practice,and to provide help for prognosis prediction and risk assessment of patients.Methods: The degree of infiltration of 434 cells in 32 cancer microenvironments was fitted by existing deep learning algorithms,and 4 cancer subtypes specific to the tumor microenvironment were systematically identified by a consistent clustering method.The differences in survival time,age,tumor stage,etc.between cancer subtype patient groups with differences in immune microenvironment were explored from the clinical characteristics of patients.To explore the differences in the degree of immune cell infiltration and immune molecular characteristics in the tumor microenvironment.The cell co-expression network method was used to explore the unique cell interaction patterns among patients with different immune microenvironment subtypes,to characterize the immune microenvironment state,and to explore the reasons for the differences in survival prognosis between patients with subtypes.A patient risk prognostic model was constructed at the cellular infiltration and transcriptomic levels.Subtype samples were further analyzed at the immune signature level to explore differences in immune signature scores,T cell-related functions,and immune checkpoint responses among pan-cancer subtypes,and finally identify a class of pan-cancer subtypes with immunotherapy potential.Results: Four immune microenvironment subtypes were identified by consistent clustering of immune cell infiltration levels across pan-cancer samples.Patients had differences in immune infiltration and survival between subtypes.Subtype 1 patients are phenotypically specific for the brain and exhibit a unique low patient age level and a higher proportion of initial state tumor stage.By screening the differences in infiltration of cell types between subtypes,we obtained cell types with specific high infiltration levels in pan-cancer subtypes 2 and 4,and further obtained subtype-specific cell interaction patterns based on cell ontology annotations.Exhibits a highly proliferative landscape of subtype 2 and a highly differentiated landscape of subtype 4.Through the risk regression model,a prognostic risk model consisting of 8cells was obtained at the cell infiltration level,and 8 genes were identified in the context of their corresponding marker gene sets as prognostic biomarkers at the transcriptome level.Subtype 4 was confirmed as a pan-cancer subtype with the potential to respond to immunotherapy through immune signature score,T cell-related gene set,and immune checkpoint-related gene signature analysis.Based on 20 differentially expressed genes between pan-cancer subtypes 2 and 4,a pan-cancer patient prognostic risk classifier was constructed.We further explored the differences in genomic mutations and copy number variation in patients with pan-cancer subtypes at the multi-omics level.High mutations in three genes can lead to poor prognosis in patients with pan-cancer subtypes.Conclusion: This study proposes a method for disease typing in cancer patients based on the tumor microenvironment cell atlas.Two prognostic models of patient risk were constructed at the cellular and transcriptomic levels by analyzing the cell types with infiltrating differences among immune subtypes and their marker genes.Biomarkers of immune subtypes were identified through transcriptome data,and a class of pan-cancer subtypes with immunotherapy potential was identified based on the performance of different immune characteristics.This study is expected to provide a new entry point for the diagnosis and treatment of patients in the actual clinical process,and to provide help for the prognosis and risk assessment of patients.