Identification of mRNA Prognostic Markers for TGCT by Integration of Co-Expression and CeRNA Network

Abstract

Testicular germ cell tumor (TGCT) is the most common malignant tumor in young men and is associated with poor prognosis. We assessed the RNA expression profiles of 13 TGCT tissues and 4 adjacent normal tissues by transcriptome sequencing to identify novel prognostic biomarkers. We detected several differentially expressed mRNAs in TGCT that were functionally annotated by GO and KEGG enrichment analyses to tumorigenesis-related processes such as immunity and chemotherapeutic resistance. An mRNA-lncRNA-miRNA regulatory network was constructed using RNA-Seq data and public databases, and integrated with TCGA database to develop a prediction model for metastasis and recurrence. Finally, GRK4, PCYT2 and RGSL1 were identified as predictive markers of survival and therapeutic response. In conclusion, we found several potential predictors for TGCT prognosis and immunotherapeutic response by ceRNA network analysis.

Keywords: chemotherapy resistance; competitive endogenous RNAs network; immune cell infiltration; prognostic markers; risk factor; testicular germ cell tumors.

Figure 1

Identification of DEmRNAs across different groups. (A-C). Volcano plots showing the DEmRNAs between (A) non-tumor vs non-SEM, (B) non-tumor vs SEM, and (C) non-tumor vs SEM and non-SEM with |Log2FC|>1 and FDR <0.01. (D) Histogram showing the DEmRNAs between groups. (E) The validation of randomly selected DEmRNAs in the GEPIA database with |Log2FC|>1 and p <0.01 as the cut-off values. *Difference was statistically significant.

Figure 2

Intergroup-specific expression and functional enrichment analysis of mRNAs. (A) The heatmap of top 100 highly expressed mRNAs in the non-tumor, SEM and non-SEM samples. (B–G). The top 20 GO terms and KEGG pathways in non-tumor (B, E), SEM (C, F), and non-SEM (D, G) groups. (H) Bubble chart of putative transcription factors of the top 100 mRNAs in three groups.

Figure 3

Enrichment analysis of common DEmRNAs. (A) Heat map showing hierarchical clustering of common DEmRNAs in the non-tumor, SEM and non-SEM groups. (B, C) Upset plots showing the distribution of (B) upregulated and (C) downregulated mRNAs for the indicated pairs. (D, E) Bar graph showing the top 20 enriched GO and KEGG pathways associated with DEmRNAs common to all three pairs. (F) The interactive network of the enriched pathways. Different colors represent different biological processes. (G) The protein-protein interaction network (PPI) of the top 100 common DEmRNAs. Color intensity is indicative of the number of connections.

Figure 4

The lncRNA-miRNA-mRNA ceRNA network in TGCT. (A, B) Wayne diagrams of DElncRNAs (A) and DEmRNAs (B) in three comparison groups, with |log2FC|>2, FDR<0.01 as the thresholds. (C) The lncRNA–miRNA–mRNA network consisting of 9 lncRNAs (blue squares), 82 miRNAs (yellow triangles) and 9 mRNAs (rose-red circles) with 166 possible interactions. (D–I) Enrichment analysis of the DEmRNAs in (D, G) non-tumor vs non-SEM, (E, H) non-tumor vs SEM, and (F, I) non-tumor vs SEM + non-SEM. The top 20 enriched GO terms and pathways are shown.

Figure 5

Development of the prognostic index based on ceRNA-related genes. (A) The coefficient of the selected feature is shown by the lambda parameter, with the value of the independent variable lambda on the horizontal axis and the coefficient of the independent variable on the vertical axis. (B) The relationship between partial likelihood deviation and log (λ) was plotted using the Lasso Cox regression model. (C) Risk score curve showing survival of patients and expression profiles of the three prognostic genes in low- and high-risk groups. (D) Kaplan-Meier curve showing the association between patient survival and the three-gene signature. Median survival duration (years) corresponds to a 50% survival rate. The ROC curve and AUC of the risk model at different times. (E–G) Survival of patients stratified based on GRK4, PCYT2, and RGSL1 expression.

Figure 6

The correlation of GRK4, PCYT2 and RGSL1 with cancer-related pathways, immune cell infiltration and GDSC. (A) Association between GSVA score and cancer-related pathways in TGCT. (B) Correlation between immune cell infiltrates and GSVA enrichment score in the selected cancers. (C) Correlation between gene expression and GDSC drug sensitivity. *P-value < 0.05; ^#FDR < 0.05.