The Correlation Between the Immune and Epithelial-Mesenchymal Transition Signatures Suggests Potential Therapeutic Targets and Prognosis Prediction Approaches in Kidney Cancer

Abstract

Both epithelial-mesenchymal transition (EMT) and immune regulation are important biological process in malignant tumours. The current research aims to comprehensively explore the potential association between the epithelial-mesenchymal transition (EMT) signature and immune checkpoint signature and its role in predicting the prognosis of clear-cell renal cell carcinoma (ccRCC) patients. EMT-related genes were collected from an experiment-based study and then were investigated using data from the Cancer Genome Atlas. A total of 357 genes were included, and 23 of them that were upregulated and correlated with prognosis were analysed further as core EMT genes in ccRCC. Interestingly, the emerging immune checkpoints CD276, OX40 and TGFB1 were found to be significantly co-expressed with core EMT genes, and TGFB1, CXCR4, IL10, and IL6 were the most important molecules potentially interacting with EMT molecules in our model, as determined from mRNA co-expression and protein-protein interaction network analysis. Additionally, an integrated scoring model based on FOXM1, TIMP1 and IL6 was successfully established to distinguish ccRCC patients with different clinical risks. Our results identified core genes in the EMT-immunophenotyping correlation and evaluated their risk assessment capabilities, providing more potential therapeutic targets and prediction approaches regarding the translational research of treatment and prognosis in ccRCC.

Figure 1

Overview of the main EMT signatures. (a) Compact visualization of 23 genomic alterations. The gene set was altered in 285 (63.9%) of 446 complete samples. Genetic alteration included amplification, deep deletion, mRNA upregulation, mRNA downregulation, truncating mutation, and missense mutation. (b) Overall survival Kaplan-Meier estimate: Cases with alterations (red): total cases: n = 285, cases deceased: n = 107, median survival in months, 74.11; cases without alterations (blue): total cases: n = 161, cases deceased: n = 43, median survival in months: NA (not available).

Figure 2

EMT signature-correlated pathways and immune checkpoint molecules. (a) Top 10 pathways enriched in the differentially expressed EMT genes. The adjusted P value is shown as dots with different colours. The enrichment count is shown as dots with different sizes. The number of candidate genes in each pathway was calculated as the GeneRatio. (b) Correlation of the expression between EMT genes and immune checkpoint genes, based on Pearson correlation coefficients.

Figure 3

Integrated network and functional annotation. (a) EMT-related genes are indicated as cyan circles, immune checkpoint targets are indicated as red circles. The thickness of the lines depends on the combined score in the PPI network. (b) A combined network contains both a protein-protein interaction network (PPI) and gene co-expression information. The thickness of the lines depends on the combination coefficient (|co-expression correlation coefficients of mRNAs* combined score in the PPI network|). (c) Functional annotation of two correlated signatures. (d) Proportion of different functional groups.

Figure 4

Expression of 15 immune checkpoint genes in 534 tumour samples and 72 adjacent normal samples. “*” indicates a log2FC value >1 and P value < 0.01.

Figure 5

Distinguishing ability of the risk score for the clinical status. (a) The Kaplan-Meier test of the risk score for overall survival. P = 8.932e-08. Prognostic performance of the risk score as shown by the time-dependent receiver operating characteristic (ROC) curve for predicting the 5-year survival. The area under the curve (AUC) = 0.701. Patients were divided into high- and low-risk groups using the risk score. The relationship between survival and the risk score is shown at the top; the risk score curve is presented in the middle; heatmap of patients with 3 signatures is on the bottom. (b) Validation of the risk score in the GEO dataset (GSE29609). The Kaplan-Meier test of the risk score for overall survival (n = 39, p value = 0.00375). Prognostic performance of the risk score as shown by the time-dependent receiver operating characteristic (ROC) curve for predicting the 5-year survival. Area under the curve (AUC) = 0.807.

Figure 6

Characteristic (ROC) curve distinguishing the different clinical parameters: grade, clinical stage, clinical M stage, clinical N stage, clinical T stage (P < 0.01).