Identification and validation a costimulatory molecule gene signature to predict the prognosis and immunotherapy response for hepatocellular carcinoma

Abstract

Background: Hepatocellular carcinoma (HCC) is one of the most common malignancies worldwide. Costimulatory molecules have been proven to be the foundation of immunotherapy. However, the potential roles of costimulatory molecule genes (CMGs) in HCC remain unclear. Our study is aimed to develop a costimulatory molecule-related gene signature that could evaluate the prognosis of HCC patients.

Methods: Based on The Cancer Gene Atlas (TCGA) database, univariate Cox regression analysis was applied in CMGs to identify prognosis-related CMGs. Consensus clustering analysis was performed to stratify HCC patients into different subtypes and compared them in OS. Subsequently, the LASSO Cox regression analysis was performed to construct the CMGs-related prognostic signature and Kaplan-Meier survival curves as well as ROC curve were used to validate the predictive capability. Then we explored the correlations of the risk signature with tumor-infiltrating immune cells, tumor mutation burden (TMB) and response to immunotherapy. The expression levels of prognosis-related CMGs were validated based on qRT-PCR and Human Protein Atlas (HPA) databases.

Results: All HCC patients were classified into two clusters based on 11 CMGs with prognosis values and cluster 2 correlated with a poorer prognosis. Next, a prognostic signature of six CMGs was constructed, which was an independent risk factor for HCC patients. Patients with low-risk score were associated with better prognosis. The correlation analysis showed that the risk signature could predict the infiltration of immune cells and immune status of the immune microenvironment in HCC. The qRT-PCR and immunohistochemical results indicated six CMGs with differential expression in HCC tissues and normal tissues.

Conclusion: In conclusion, our CMGs-related risk signature could be used as a prediction tool in survival assessment and immunotherapy for HCC patients.

Keywords: Costimulatory molecule; Hepatocellular carcinoma; IPS; Prognostic signature; ssGSEA.

Fig. 1

The flowchart of the study

Fig. 2

Expressions of the CMGs in HCC. A The expression levels of 59 CMGS in both HCC tissues and normal samples. *P < 0.05, **P < 0.01, ***P < 0.001. B The correlation of the 59 CMGs by using Spearman correlation analysis. C PPI network showed the interactions of the CMGs (the highest confidence: 0.9)

Fig. 3

Consensus clustering analysis for HCC patients based on the CMGs. A Univariate Cox regression analysis identified prognosis-related CMGs. B Consensus clustering matrix for k = 2. C Principal Component Analysis (PCA) plot for clusters. D Kaplan–Meier overall survival (OS) curves of Cluser1 and Cluster 2. E Heatmap and clinical factors of the two clusters

Fig. 4

Construction of CMGs risk signature for HCC. A The distribution of the risk score, B survival status, C expression of 6 prognosis-related CMGs in high- and low-risk groups, D Kaplan–Meier survival curve, E time-dependent ROC curve analyses of the CMGs risk signature in the training set. F The distribution of the risk score, G survival status, H expression of 6 prognosis-related CMGs in high- and low-risk groups, I Kaplan–Meier survival curve, J time-dependent ROC curve analyses of the CMGs risk signature in the test set. K The distribution of the risk score, L survival status, M expression of 6 prognosis-related CMGs in high- and low-risk groups, N Kaplan–Meier survival curve, O time-dependent ROC curve analyses of the CMGs risk signature in the total set

Fig. 5

Univariate and multivariate Cox regression analyses. Univariate and Multivariate Cox regression analysis of the correlation between the risk score and clinicopathological features in the training set (A, B), test set (C, D) and total set (E, F), respectively

Fig. 6

Clinical characteristics of CMGs prognostic signature in different subgroups. A The heatmap and clinicopathological factors of high- and low-risk subgroups. *P < 0.05, **P < 0.01, ***P < 0.001. B The student’s t-test was used to assessed the relationship between the CMGs prognostic signature and age, gender, stage, grade, TMN stage. C Kaplan–Meier survival analyses of CMGs risk model in different clinical subgroups based on age, gender, stage, grade, TMN stage with log-rank test

Fig. 7

Construction and validation of a novel nomogram. A The nomogram for predicting 1-year, 3-year and 5-year OS of HCC in total set. B The calibration curves for internal validation of the nomogram on consistency between predicted and observed 1-year, 3-year and 5-year OS in total set. C The time-dependent ROC of the nomogram and clinical factors for 1-year, 3-year and 5 year OS prediction in total set

Fig. 8

Analysis of tumor infiltrating immune cells. A The association of CMGs prognostic signature and immune cells infiltration. Significant statistical differences between two risk groups were assessed by the Wilcoxon rank-sum test, *P < 0.05, **P < 0.01, ***P < 0.001. B The relationship between OS and immune cells infiltration (naïve B cells, resting memory CD4 T cells, activated memory CD4 T cells, regulatory T cells, gamma delta T cells, macrophage M1 and resting mast cells)

Fig. 9

Comparison of the immune status in high- and low-risk groups. A The immune status of HCC patients in high- and low-risk groups. Tumor purity, ESTIMATE score, immune score and stromal score of every sample were showed in the heatmap. B The box plot displayed the differences of enrichment scores of 16 types of immune cells and 13 immune-related pathways in high- and low-risk groups using Mann–Whitney test. C The differences of stromal score, immune score, ESTIMATE score and tumor purity in high- and -risk groups with violin plots. *** P < 0.001

Fig. 10

Evaluation of tumor mutation burden and the response to immunotherapy among high- and low-risk groups. A Mutation profiles of high- and low-risk groups. B The relationship between CMGs risk signature and TMB. C The association between IPS and risk signature for HCC patients

Fig. 11

Verification of prognostic genes in HCC tissues and normal tissues. A TNFSF4, B TNFRSF4, C TNFRSF11A, D TNFRSF11B, E TMIGD2, F CD40LG. ns: not significant,*P < 0.05,**P < 0.01

Fig. 12

Representative immunohistochemistry images of A TNFSF4, B TNFRSF4, C TNFRSF11A, D TNFRSF11B, E TMIGD2, F CD40LG in HCC and normal liver tissues derived from the HPA database. HCC hepatocellular carcinoma, HPA Human Protein Atlas