Abstract
BACKGROUND: The global burden of hepatocellular carcinoma (HCC) is increasing, negatively impacting social health and economies. The discovery of novel and valuable biomarkers for the early diagnosis and therapeutic guidance of HCC is urgently needed. METHODS: Extracellular matrix (ECM)-related gene sets, transcriptome data and mutation profiles were downloaded from the Matrisome Project and The Cancer Genome Atlas (TCGA)-LIHC datasets. Coexpression analysis was initially performed with the aim of identifying ECM-related lncRNAs (r > 0.4, p < 0.001). The screened lncRNAs were subjected to univariate analysis to obtain a series of prognosis-related lncRNA sets, which were incorporated into least absolute selection and shrinkage operator (LASSO) regression for signature establishment. Following the grouping of LIHC samples according to risk score, the correlations between the signature and clinicopathological, tumour immune infiltration, and mutational characteristics as well as therapeutic response were also analysed. lncRNA expression levels used for modelling were finally examined at the cellular and tissue levels by real-time PCR. All analyses were based on R software. RESULTS: AL031985.3 and MKLN1-AS were ultimately identified as signature-related lncRNAs, and both were significantly upregulated in HCC tissue samples and cell lines. The prognostic value of the signature reflected by the AUC value was superior to that of age, sex, grade and stage. Correlation analysis results demonstrated that high-risk groups exhibited significant enrichment of immune cells (DCs, macrophages and Tregs) and increased expression levels of all immune checkpoint genes. Prominent differences in clinicopathological profiles, immune functions, tumour mutation burden (TMB) and drug sensitivity were noted between the two risk groups. CONCLUSIONS: Our signature represents a valuable predictive tool in the prognostic management of HCC patients. Further validation of the mechanisms involved is needed.