Identification of a Costimulatory Molecule-Related Signature for Predicting Prognostic Risk in Prostate Cancer

Abstract

Costimulatory molecules have been proven to enhance antitumor immune responses, but their roles in prostate cancer (PCa) remain unexplored. In this study, we aimed to explore the gene expression profiles of costimulatory molecule genes in PCa and construct a prognostic signature to improve treatment decision making and clinical outcomes. Five prognosis-related costimulatory molecule genes (RELT, TNFRSF25, EDA2R, TNFSF18, and TNFSF10) were identified, and a prognostic signature was constructed based on these five genes. This signature was an independent prognostic factor according to multivariate Cox regression analysis; it could stratify PCa patients into two subgroups with different prognoses and was highly associated with clinical features. The prognostic significance of the signature was well validated in four different independent external datasets. Moreover, patients identified as high risk based on our prognostic signature exhibited a high mutation frequency, a high level of immune cell infiltration and an immunosuppressive microenvironment. Therefore, our signature could provide clinicians with prognosis predictions and help guide treatment for PCa patients.

Keywords: bioinformatics; biomarker; costimulatory molecules; prognostic signature; prostate cancer.

FIGURE 1

The flowchart of the present study design.

FIGURE 2

The Kaplan–Meier curves for the 14 costimulatory molecule genes in prostate cancer from The Cancer Genome Atlas dataset, including TNFSF10, TNFSF18, TNFRSF6B, TNFRSF18, TNFRSF25, CD80, CD86, CD70, RELT, EDA2R, LTA, CD276, TNFSF13, and LTBR.

FIGURE 3

Consensus clustering based on the five costimulatory molecule genes. (A) Consensus clustering cumulative distribution function (CDF) for k = 2 to k = 10. (B) The relative change in area under the CDF curve for k = 2 to k = 10. (C) The Kaplan–Meier curve evaluate the prognosis of prostate cancer patients. (D) The Gene Set Enrichment Analysis showed that several oncogenic pathways were significantly enriched in cluster 1.

FIGURE 4

The association between the expression levels of the five costimulatory molecules and clinical factors. (A) The expression levels of the five costimulatory molecules between the normal and tumor samples; (B) T2 staging vs. T3 and T4 staging; (C) N0 staging vs. N1 staging. (D) Gleason score for 6, 7, 8, 9, and 10. GS represents the Gleason score. The unpaired Student’s t-test was performed for comparison between two samples and the one-way analysis of variance test for comparison between multiple samples.

FIGURE 5

Construction and validation of a prognostic-related risk score model. The Kaplan–Meier curves, time-dependent receiver operating characteristic curves and principal component analysis for The Cancer Genome Atlas dataset (A–C), GSE21034 dataset (D–F), GSE54460 dataset (G–I), GSE70768 dataset (J–L), and GSE70769 dataset (M–O).

FIGURE 6

Relationship between the prognostic signature and clinicopathological factors of PCa patients. (A) The heat map shows the expressions of the five costimulatory molecule genes and clinicopathological factors in the high- and low-risk groups. The univariate (B) and multivariate (C) Cox regression analyses of clinicopathological factors (including the risk score) and prognosis. The bar chat showed that the prognostic signature had significantly different in different clinical subgroups, and the PCa patients with advanced age (D), high pathological T stage (E), node metastasis (F), high Gleason score (G), high prostate-specific antigen (H), and recurrence (I) tend to have a high risk score. ns: P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

FIGURE 7

The costimulatory molecule-based signature-related biological pathways. (A) The most related genes of costimulatory molecule-based signature in PCa (Pearson | R| > 0.4). (B–E) The GO and KEGG analysis of the identified the potential functions and pathways of costimulatory molecule genes.

FIGURE 8

Immune microenvironment and tumor mutation burden in high-risk and low-risk patients. (A) The gene expression profile of 28 immune cell types in high- and low-risk patients. (B) Box plots showing 28 differential immune cell infiltration difference between high- and low-risk patients. (C–E) The immune score, stromal score and ESTIMATE score in high- and low-risk patients. (F) Patients in high-risk group had higher tumor mutation burden than those in low-risk group. (G,H) The mutation profile of the top 20 mutation genes in the low- and high-risk groups. (I) Forest plot illustrated the differences of mutation frequency of several gene (TP53, STAB2, MUC17, PCDHB7, CUBN, CACNA1A, and MXRA5) in high- and low-risk patients. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.