Development of multivariable risk signature based on four immune-related RNA-binding proteins to predict survival and immune status in lung adenocarcinoma

Abstract

Background: This study aimed to construct a risk signature with predictive power based on immune-related RNA-binding proteins (RBPs) for lung adenocarcinoma (LUAD) patients.

Methods: The Cancer Genome Atlas (TCGA) database was used as the data source. Immune genes (IGs) were obtained from the Immunology Database and Analysis Portal (immPort) database. Differentially expressed RBPs and IGs between tumor and normal tissues were screened. For external validation, an independent cohort from the Gene-Expression Omnibus (GEO) database was used. The accuracy of the risk signature prediction was evaluated using Cox regression analysis and the receiver operating characteristic (ROC) curve.

Results: The risk signature was constructed from four immune-related and prognostic RBPs (OAS3, PCF11, TLR7, and EXO1). The patients were divided into the low- and high-risk groups, with the low-risk group having a higher survival rate than the high-risk group. The risk signature outperformed other clinical parameters, with a multivariable hazard ratio of 1.862 (95% confidence interval: 1.292-2.683). The tumor immune microenvironment, stemness index, immune checkpoint, immune infiltration, and proportion of immune cells were significantly different between the low- and high-risk groups (all P<0.05).

Conclusions: The risk signature of immune-related RBPs can provide the basis for clinical decisions regarding diagnosis, prognosis, and immunotherapy in LUAD patients.

Keywords: Lung adenocarcinoma (LUAD); RNA-binding protein (RBP); drug sensitivity; immune infiltration; prognostic model.

Figure 1

Differentially expressed genes in LUAD patients. (A) Heatmap and volcano plot of differentially expressed RBPs in LUAD patients. (B) Heatmap and volcano plot of differentially expressed IGs in LUAD patients. (C) Regulatory network and functional enrichment analysis. The regulatory network of RBPs and IGs. GO (D) and KEGG (E) enrichment analyses of the immune-related RBPs. (F) Univariable Cox regression analyses. (G) Multivariable Cox regression analysis. LUAD, lung adenocarcinoma; RBP, RNA-binding protein; IG, immune gene; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; FC, fold change; ncRNA, non-coding RNA; UTR, untranslated region; BP, biological process; CC, cellular component; MF, molecular function; FDR, false discovery rate.

Figure 2

Construction of the prognostic model in the TCGA cohort based on risk scores. (A) Overall survival of the low- and high-risk groups. (B) ROC curve of the TCGA cohort. (C) The risk score distribution and survival status of LUAD patients. (D) Heat map of risk genes expression. AUC, area under the curve; TCGA, The Cancer Genome Atlas; ROC, receiver operating characteristic; LUAD, lung adenocarcinoma.

Figure 3

Validation of the prognostic model in the GSE68465 cohort based on risk scores. (A) Overall survival of the low- and high-risk groups. (B) ROC curve of the GSE68465 cohort. (C) The risk score distribution and survival status of LUAD patients. (D) Heat map of risk genes expression. (E) The association between risk score and clinical parameters (age, gender, stage, T, N, and M). AUC, area under the curve; ROC, receiver operating characteristic; LUAD, lung adenocarcinoma.

Figure 4

Independent prognostic analysis and nomogram for prediction. Univariable (A) and multivariable (B) cox regression analysis. (C) Comparison of risk signature and clinical parameters in predicting the prognosis of LUAD. (D) Development of a nomogram for predicting 1-, 3-, and 5-year OS of LUAD patients. AUC, area under the curve; LUAD, lung adenocarcinoma; OS, overall survival.

Figure 5

Analysis of risk score with the tumor immune microenvironment, stemness index, and immune checkpoint in LUAD. (A) Based on the ESTIMATE algorithm, correlation analysis risk score with stromal cell scores, immune cell scores, DNAss, and RNAss. (B) R means correlation value. The immune checkpoint gene expression, including CTLA4, LAG3, TIGIT, and TIM3, in the low- and high-risk groups. (C) Immune cell infiltration analysis. Correlations between risk signature and immune cell infiltration levels, including B cell, CD4⁺ T cell, CD8⁺ T cell, dendritic cells, neutrophil, macrophage. LUAD, lung adenocarcinoma; DNAss, DNA stemness score; RNAss, RNA stemness score.

Figure 6

Distribution pattern and gene enrichment of low- and high-risk groups. (A) Composition of 22 immune cells in low- and high-risk groups. (B) Correlation heat map of 22 immune cells in LUAD. (C) PCA for low- and high-risk groups based on risk genes, whole RBPs, whole IGs, and whole genome. (D) GSEA suggested differences in immune response and immune system process. (E) GSEA suggested differences in the KEGG pathway between the low- and high-risk groups. LUAD, lung adenocarcinoma; PCA, Principal components analysis; RBP, RNA-binding protein; IG, immune gene; GSEA, gene set enrichment analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Figure 7

Drug sensitive to risk genes. (A) Alectinib. (B) Denileukin diftitox (ontak). (C) Irofulven. (D) Fluphenazine. (E) Isotretinoin. (F) LDK-378. (G) Imiquimod. (H) Vorinostat. (I) Megestrol acetate. (J) Nelarabine. (K) Nelfinavir. (L) Estramustine. (M) 6-thioguanine. (N) Vorinostat. (O) Celecoxib. (P) Dromostanolone propionate.