Exploring Extracellular Vesicles Data To Identify Prognostic Gene Signatures For Lung Adenocarcinoma

Background:Globally,lung cancer is the second most common cancer after breast cancer and the leading cause of cancer deaths.Non-small cell lung cancer represents more than 80 to 85% of lung cancers,of which approximately50% are lung adenocarcinoma(LUAD).The 5-year survival rate for LUAD is only 19%.Extracellular vesicles(EV)are nano-sized lipid bilayer vesicles of endocytic origin,containing nucleic acids,proteins,and lipids.EV plays decisive roles in cell-to-cell communication.By transferring oncogenic proteins and nucleic acids,EVs play an important role in modulating tumorigenesis,growth,and metastasis,as well as,they are also novel biomarkers for the diagnosis and prognosis of various cancers.However,it is still need to clarify how to use EV-related genes to predict the outcome in LUAD patients.The traditional tumor staging system and histopathologic characterization are insufficient for precise prediction of the outcomes for LUAD patients;therefore,more accurate prognostic tools are needed by clinicians to optimize individual therapeutic strategies.Objective:To explore an EV-related prognostic gene signature for LUAD,and construct a more accurate prognostic model for clinicians to optimize individual therapeutic strategies.Methods:1.Generation of a prognostic signature in the TCGA-LUAD cohort1)Homo species EV-associated genes were chosen from the ExoCarta database.Both RNA sequencing and clinical characteristics data for LUAD patients were downloaded from the TCGA database.EV-associated genes differentially expressed in tumor tissues and adjacent normal tissues in the TCGA-LUAD cohort were identified using the "limma" package.Univariate Cox regression analysis was adopted to screen the EV-related genes whose expression was significantly correlated with overall survival(OS).2)The corresponding multivariable prognosis prediction model was constructed using the LASSO Cox regression method.A signature-based risk score for each case was computed with the gene expression level and regression coefficient.3)Using the median risk score as the cutoff value,patients were classified into low or high-risk groups.Kaplan–Meier and time‐dependent ROC curve analysis was conducted to evaluate the predictive power of the gene signature.2.Prognostic gene signature was validated in GEO-LUAD cohort1)Microarray and clinical data for 442 LUAD tumor samples were extracted from the GSE68465 dataset in the GEO database.2)The patients were classified into high and low-risk groups as previously prediction model.3)Kaplan–Meier and time‐dependent ROC curve analysis was conducted to evaluate the predictive power of the gene signature.3.Variation analysis in the TCGA-LUAD cohort1)GO enrichment was performed to explore the molecular function,cellular component,and biological process of the differentially expressed genes(DEGs)among the low and high-risk groups LUAD.2)KEGG analyses were performed to explore the pathway enrichment.3)The ssGSEA scores for 16 immune cells and 13 immune-related pathways were calculated to evaluate immune infiltration in LUAD tissues.Result:1.Generation of a prognostic signature in the TCGA-LUAD cohort1)A total of 720 homo species EV-associated genes were chosen from the ExoCarta database.2)The 83 EV-associated genes defined as DEGs by “limma” were expressed differently in LUAD tissues and the adjacent normal tissues.The 81 genes associated with OS were identified with univariate Cox regression analysis.A total of 52 DEGs were significantly associated with OS.3)we constructed a prognostic prediction gene signature with LASSO Cox regression,which containing 19 EV-associated genes.The linear combination of the m RNA expression level for each EV-associated gene was used to compute a risk score and patients were classified into low or high-risk groups based on the median risk score.4)Kaplan-Meier cumulative curves revealed the OS was significantly longer in the low-risk group.The risk score was an independent prognostic predictor of OS in the multivariate analysis.ROC curve analyses revealed the gene signature had high specificity and specificity for the OS.The AUC values were: 1 year,0.720;2 years,0.728;and 3 years,0.735.2.Prognostic gene signature was validated in GEO LUAD cohort1)The patients of GSE68465 were classified into high and low-risk groups as previously described.2)Kaplan-Meier cumulative curves revealed the OS was significantly longer in the low-risk group,multivariate analysis also showed the risk score was an independent prognostic predictor for OS.Time-dependent ROC curve analyses revealed the 19-gene signature-based risk score had high specificity and specificity for the OS in LUAD patients from GSE68465.The AUC values were: 1 year,0.686;2 years,0.645,and 3 years,0.627.3.Enrichment analysis in the TCGA-LUAD cohortKEGG analysis indicated that the DEGs among the two groups were mostly enriched in “extracellular matrix(ECM)-receptor interaction”.GO analysis indicated these genes were mostly enriched in several immunity processes and molecular functions.4.Immune analysis in the TCGA-LUAD cohortObvious discrepancies in immune cells and immune functions between these two groups were identified using ssGSEA analysis.Conclusion:1.EV-related genes were related to tumorigenesis,growth,and metastasis.2.We constructed a novel 19 EV-related gene prognostic model.3.EVs play a key role in ECM degradation and remodeling,as well as,increase glycolysis inducing a hypoxic state,and promoted tumor progression.4.EVs are critical mediators of the interactions between cancer and immune cells and regulate anti-and pro-tumorigenic functions.