| Purpose:Plenty of long non-coding RNAs contributes to regulate immune response,subsequently adjusting tumor development.This study aims to investigate a novel classification for non-small cell lung cancer in accordance with expression of long non-coding RNA and thereupon establish an immune-related long non-coding RNA signature as a prognostic model.Method:RNA sequence from patients with resectable non-small cell lung cancer(stage Ⅰ-ⅢB)are downloaded in The Cancer Genome Atlas website and Gene Expression Omnibus database.Immune-related long non-coding RNAs in non-small cell lung cancer are acquired from Immlnc database to identify crossing long non-coding RNAs in three databases,which are filtrated by univariate cox analysis.Novel classification is investigated through unsupervised clustering and Kaplan-Meier survival analysis.Lasso and multivariate cox analysis are conducted to establish prognostic model incorporated with American Joint Committee on Cancer Staging Manual 8th edition,which is evaluated by Kaplan-Meier method and receiver characteristic operator curve.Result:From TCGA,970 patients were assigned into training and internal validation cohorts as 6:4.Gene Expression Omnibus datasets with 396 patients composed external validation cohort.Univariate analysis identified 1,034 immune-related long non-coding RNA in non-small cell lung cancer(P<0.01).Among different cluster divisions,unsupervised clustering with 3 division shows most significant survival difference.Targeting tumor microenvironment has unprecedented predicted effect for immunotherapy over immune checkpoints and tumor mutation burden(TMB).Long non-coding RNA(LncRNA)emerges as a critical regulator of immune system by modifying immune-related genes in epigenetic,transcriptional,and post-transcriptional processes.We constructed three clusters based on immune-related IncRNAs,of which the immune contextures in tumor microenvironment are consistent with immune-inflammed,immune-evaded and immune-desert phenotypes.Clusters were interpreted within individual tumors that could predict landscape of tumor infiltrating immune cells,immune and pathologic subtypes,functional pathways,genetic variation and patient prognosis.Cluster A was characterized by the immune-activating signaling pathways,while activation of interferon-gamma and MHC class I signaling pathways could be detected in cluster B.And cluster C was enriched in cell adhesion.Somatic variation disclosed 31%mutation rate of KRAS in cluster A,25%mutation rate of KEAP1 in cluster B and 80%mutation rate of TP53 in cluster C.Increased tumor mutation burden(TMB)were observed in cluster B.In terms of therapeutic sensitivity,cluster C had better predicted response for chemotherapy.Reciprocally,cluster A and B are more sensitive to target therapy.TIDE predicted favorable response for anti-PD-1/L1 immunotherapy in cluster A,which could be reasoned by highest expression of PD-1 and CTLA-4.Optimal long non-coding RNAs identified by LASSO were CDC42.IT1,LINC01116,WWC2.AS2,CASC15,BANCR,COLCA1,PRKG1.AS1,RPARP.AS1,HOTAIR,TMPO.AS1,CRNDE,C20orf197,CSNK1G2.AS1,LINC00528,LINC0089.Models established by multivariate analysis exhibited area under the curve of 0.69 in training cohort,0.61 in internal validation cohort and 0.61 in external validation cohort,with significant Kaplan-Meier analyses among three cohorts(P<0.001).American Joint Committee on Cancer Staging Manual 8th edition is attached to the model,showing an area under the curve of 0.90 in training cohort,0.92 in internal validation cohort and 0.63 in external validation cohort.Conclusion:Significant survival-related clustering of long non-coding RNAs could be applied into novel classification for non-small cell lung cancer.The differences in immune microenvironment of these three clusters in non-small cell lung cancer were analyzed in terms of genomic characteristics,immune checkpoint,TMB,chemotherapy,targeted therapy and immunotherapy,suggesting that targeted immune microenvironment is better than simple immune checkpoint or TMB in immunotherapy.Immune-related long non-coding RNAs are independent prognostic factors and display better predictive power incorporating American Joint Committee on Cancer Staging,which indicates potential clinical impaction. |
PDF Full Text Request