A prognostic model of non small cell lung cancer based on TCGA and ImmPort databases

Abstract

Bioinformatics methods are used to construct an immune gene prognosis assessment model for patients with non-small cell lung cancer (NSCLC), and to screen biomarkers that affect the occurrence and prognosis of NSCLC. The transcriptomic data and clinicopathological data of NSCLC and cancer-adjacent normal tissues were downloaded from the Cancer Genome Atlas (TCGA) database and the immune-related genes were obtained from the IMMPORT database ( http://www.immport.org/ ); then, the differentially expressed immune genes were screened out. Based on these genes, an immune gene prognosis model was constructed. The Cox proportional hazards regression model was used for univariate and multivariate analyses. Further, the correlations among the risk score, clinicopathological characteristics, tumor microenvironment, and the prognosis of NSCLC were analyzed. A total of 193 differentially expressed immune genes related to NSCLC were screened based on the "wilcox.test" in R language, and Cox single factor analysis showed that 19 differentially expressed immune genes were associated with the prognosis of NSCLC (P < 0.05). After including 19 differentially expressed immune genes with P < 0.05 into the Cox multivariate analysis, an immune gene prognosis model of NSCLC was constructed (it included 13 differentially expressed immune genes). Based on the risk score, the samples were divided into the high-risk and low-risk groups. The Kaplan-Meier survival curve results showed that the 5-year overall survival rate in the high-risk group was 32.4%, and the 5-year overall survival rate in the low-risk group was 53.7%. The receiver operating characteristic model curve confirmed that the prediction model had a certain accuracy (AUC = 0.673). After incorporating multiple variables into the Cox regression analysis, the results showed that the immune gene prognostic risk score was an independent predictor of the prognosis of NSCLC patients. There was a certain correlation between the risk score and degree of neutrophil infiltration in the tumor microenvironment. The NSCLC immune gene prognosis assessment model was constructed based on bioinformatics methods, and it can be used to calculate the prognostic risk score of NSCLC patients. Further, this model is expected to provide help for clinical judgment of the prognosis of NSCLC patients.

Figure 1

A heat map (A) and a volcano map (B) of differential immune genes. The heat map abscissa represents the sample: the blue area represents normal tissue and the red area represents tumor tissue; the ordinate represents the gene. On the volcano map, the green area represents the downregulated differential genes and the red area represents the upregulated differential genes.

Figure 2

Heat map and volcano map of differentially expressed TFs of non-small cell lung cancer. A heat map of differentially expressed TFs of non-small cell lung cancer, red represents high expression, blue represents low expression; B volcano map of TFs of non-small cell lung cancer. The X-axis is log FC, and the larger the absolute value is, the larger the corrected P value is, indicating the larger the multiple of the difference is. The Y-axis is the corrected P value, and the larger the logarithm of log10 is, indicating the more significant the difference is.

Figure 3

Transcriptional factors and immune gene regulatory network (Triangles represent transcription factors, circles represent high-risk immune genes, and cones represent low-risk immune genes; The red line represents positive regulation, and the blue line represents negative regulation).

Figure 4

Forest map of 19 differentially expressed immune genes in the univariate Cox regression model.

Figure 5

Kaplan–Meier survival analysis of non-small-cell lung cancer patients by risk stratification.

Figure 6

Risk score curve and survival heat map. (A) survival heat map, with the increase of risk score, the expression of immune genes increased; (B) risk score curve, from left to right, the patient's risk score increased gradually; (C) point of survival chart (With the increase of patients'risk value, more patients died).

Figure 7

Principal component analysis plot using expression values at 13 selected immune genes.

Figure 8

ROC curve of multivariate Cox analysis model.

Figure 9

Immune gene prediction model (decision tree algorithm).

Figure 10

ROC curve of decision tree algorithm model.

Figure 11

Cox multivariate regression analysis.

Figure 12

ROC curve of multivariate Cox analysis model in GSE68465 database.

Figure 13

ROC curve of multivariate Cox analysis model in GSE101929 database.

Figure 14

Correlation analysis between risk score and immune cell infiltration in tumor microenvironment.