Abstract
OBJECTIVE: Vascular invasion (VI) profoundly impacts the prognosis of hepatocellular carcinoma (HCC), yet the underlying biomarkers and mechanisms remain elusive. This study aimed to identify prognostic biomarkers for HCC patients with VI. METHODS: Transcriptome data from primary HCC tissues and HCC tissues with VI were obtained through the Genome Expression Omnibus database. Differentially expressed genes (DEGs) in the two types of tissues were analyzed using functional enrichment analysis to evaluate their biological functions. We examined the correlation between DEGs and prognosis by combining HCC transcriptome data and clinical information from The Cancer Genome Atlas database. Univariate and multivariate Cox regression analyses, along with the least absolute shrinkage and selection operator (LASSO) method were utilized to develop a prognostic model. The effectiveness of the model was assessed through time-dependent receiver operating characteristic (ROC) curve, calibration diagram, and decision curve analysis. RESULTS: In the GSE20017 and GSE5093 datasets, a total of 83 DEGs were identified. Gene Ontology analysis indicated that these DEGs were predominantly associated with xenobiotic stimulus, collagen-containing extracellular matrix, and oxygen binding. Additionally, Kyoto Encyclopedia of Genes and Genomes analysis revealed that the DEGs were primarily involved in immune defense and cellular signal transduction. Cox and LASSO regression further identified 7 genes (HSPA8, ABCF2, EAF1, MARCO, EPS8L3, PLA3G1B, C6), which were used to construct a predictive model in the training cohort. We used X-tile software to calculate the optimal cut-off value to stratify HCC patients into low-risk and high-risk groups. Notably, the high-risk group exhibited poorer prognosis than the low-risk group (P < 0.001). The model demonstrated area under the ROC curve (AUC) values of 0.815, 0.730, and 0.710 at 1-year, 3-year, and 5-year intervals in the training cohort, respectively. In the validation cohort, the corresponding AUC values were 0.701, 0.571, and 0.575, respectively. The C-index of the calibration curve for the training and validation cohorts were 0.716 and 0.665. Decision curve analysis revealed the model's efficacy in guiding clinical decision-making. CONCLUSIONS: The study indicates that 7 genes may be potential prognostic biomarkers and treatment targets for HCC patients with VI.