Clinical Significance of a Novel Vasculogenic Mimicry-based Prognostic Model in Hepatocellular Carcinoma

Abstract

Background: Vasculogenic mimicry, a novel neovascularization pattern of aggressive tumors, is associated with poor clinical outcomes.

Objective: The aim of this research was to establish a new model, termed VC score, to predict the prognosis, Tumor Microenvironment (TME) components, and immunotherapeutic response in Hepatocellular Carcinoma (HCC).

Methods: The expression data of the public databases were used to develop the prognostic model. Consensus clustering was performed to confirm the molecular subtypes with ideal clustering efficacy. The high- and low-risk groups were stratified utilizing the VC score. Various methodologies, including survival analysis, single-sample Gene Set Enrichment Analysis (ssGSEA), Tumor Immune Dysfunction and Exclusion scores (TIDE), Immunophenoscore (IPS), and nomogram, were utilized for verification of the model performance and to characterize the immune status of HCC tissues. GSEA was performed to mine functional pathway information.

Results: The survival and immune characteristics varied between the three molecular subtypes. A five-gene signature (TPX2, CDC20, CFHR4, SPP1, and NQO1) was verified to function as an independent predictive factor for the prognosis of patients with HCC. The high-risk group exhibited lower Overall Survival (OS) rates and higher mortality rates in comparison to the low-risk group. Patients in the low-risk group were predicted to benefit from immune checkpoint inhibitor therapy and exhibit increased sensitivity to immunotherapy. Enrichment analysis revealed that signaling pathways linked to the cell cycle and DNA replication processes exhibited enrichment in the high-risk group.

Conclusion: The VC score holds the potential to establish individualized treatment plans and clinical management strategies for patients with HCC.

Keywords: Hepatocellular carcinoma; immunotherapy.; machine learning; prognostic prediction model; tumor microenvironment; vasculogenic mimicry.

Fig (1)

Consensus clustering of patients with Hepatocellular Carcinoma (HCC) based on Vasculogenic Mimicry (VM)-related genes. (A) The intersection of Differentially Expressed Genes (DEGs) identified from the Cancer Genome Atlas-Liver Hepatocellular Carcinoma (TCGA-LIHC) cohort with VM-related genes. (B) 22 prognosis-related genes. (C) K-M survival analysis of TCGA-LIHC cohort. (D) K-M survival analysis of the GSE14520 cohort.

Fig (2)

Immune and signaling pathway characteristics of the three molecular subtypes. (A) Differential gene expression levels and clinical characteristics between molecular subtypes. (B) Estimation of Stromal and Immune cells in Malignant Tumors using Expression data (ESTIMATE) scores. (C) Innate immunity and adaptive immunity scores. (D) Comparison of 27 immune components evaluated using single-sample Gene Set Enrichment Analysis (ssGSEA). (E) Gene Set Enrichment Analysis (GSEA) of differentially expressed genes between the molecular subtypes.

Fig (3)

Establishment and validation of the risk model (VC score). (A) Venn diagram of Differentially Expressed Genes (DEGs). (B) Least absolute shrinkage and selection operator and Cox regression analyses were executed to reduce the number of genes. (C-K) Receiver Operating Characteristic (ROC) curve, K-M survival curves, and survival status of TCGA-LIHC (C-E), International Cancer Genome Consortium-Japanese Liver Cancer (ICGC-LIRI-JP) (F-H), and GSE14520 (I-K) cohorts.

Fig (4)

Construction and identification of nomogram. (A-B). Results of univariate and multivariate Cox regression analyses. (C) Nomogram constructed with American Joint Committee on Cancer (AJCC) stage and VC score. (D) The calibration curve of the nomogram model.

Fig (5)

Prediction of differential immune landscapes across high and low-risk groups. (A) Scatter plot of the correlation between VC score and Epithelial-mesenchymal Transition (EMT), Cancer Stem Cells (CSCs), and tumor proliferation rate. (B) The immune score was derived through the ESTIMATE algorithm. (C) Innate immunity and adaptive immunity scores. (D) Comparison of 27 immune components.

Fig (6)

Prediction of immunotherapy response using the VC score. (A-B). Differential Tumor Immune Dysfunction and Exclusion (TIDE) scores and response ratios. (C-D) The fan diagram shows the differential Immunophenoscores (IPSs) between high- and low-risk groups. (E-F) ROC and K-M curves of the IMvigor210 cohort. (G-H) The relationship between VC score and efficacy of immunotherapy. Abbreviations: CR, complete response; PR, partial response; SD, stable disease; PD, progressive disease.

Fig 7

Enrichment of biological functions and the results of quantitative Real-time Polymerase Chain Reaction (qRT-PCR). (A-B) The results of GSEA in different risk subgroups. (C) The expression levels of five signature genes that were used to construct the VC score.