ABSTRACT
Background: Due to the heterogeneity of Hepatocellular carcinoma (HCC), there is an urgent need for reliable diagnosis and prognosis. Mitochondria-mediated abnormal lipid metabolism affects the occurrence and progression of HCC. Objective: This study aims to investigate the potential of mitochondrial lipid metabolism (MTLM) genes as diagnostic and independent prognostic biomarkers for HCC. Methods: MTLM genes were screened from the Gene Expression Omnibus (GEO) and Gene Set Enrichment Analysis (GSEA) databases, followed by an evaluation of their diagnostic values in both The Cancer Genome Atlas Program (TCGA) and the Affiliated Cancer Hospital of Guangxi Medical University (GXMU) cohort. The TCGA dataset was utilized to construct a gene signature and investigate the prognostic significance, immune infiltration, and copy number alterations. The validity of the prognostic signature was confirmed through GEO, International Cancer Genome Consortium (ICGC), and GXMU cohorts. Results: The diagnostic receiver operating characteristic (ROC) curve revealed that eight MTLM genes have excellent diagnostic of HCC. A prognostic signature comprising 5 MTLM genes with robust predictive value was constructed using the lasso regression algorithm based on TCGA data. The results of the Stepwise regression model showed that the combination of signature and routine clinical parameters had a higher area under the curve (AUC) compared to a single risk score. Further, a nomogram was constructed to predict the survival probability of HCC, and the calibration curves demonstrated a perfect predictive ability. Finally, the risk score also unveiled the different immune and mutation statuses between the two different risk groups. Conclusion: MTLT-related genes may serve as diagnostic and prognostic biomarkers for HCC as well as novel therapeutic targets, which may be beneficial for facilitating further understanding the molecular pathogenesis and providing potential therapeutic strategies for HCC.
ABSTRACT
Background: Hepatocellular carcinoma (HCC), the main type of liver cancer, is the second most lethal tumor worldwide, with a 5-year survival rate of only 18%. Driver genes facilitate cancer cell growth and spread in the tumor microenvironment. Here, a comprehensive driver gene signature for the prognosis of HCC was developed. Methods: HCC driver genes were analyzed comprehensively to develop a better prognostic signature. The dataset of HCC patients included mRNA sequencing data and clinical information from the TCGA, the ICGC, and the Guangxi Medical University Cancer Hospital cohorts. First, LASSO was performed to develop a prognostic signature for differentially expressed driver genes in the TCGA cohort. Then, the robustness of the signature was assessed using survival and time-dependent ROC curves. Furthermore, independent predictors were determined using univariate and multivariate Cox regression analyses. Stepwise multi-Cox regression analysis was employed to identify significant variables for the construction of a nomogram that predicts survival rates. Functional analysis by Spearman correlation analysis, enrichment analysis (GO, KEGG, and GSEA), and immunoassay (ssGSEA and xCell) were performed. Result: A 4-driver gene signature (CLTC, DNMT3A, GMPS, and NRAS) was successfully constructed and showed excellent predictive efficiency in three cohorts. The nomogram indicated high predictive accuracy for the 1-, 3-, and 5-year prognoses of HCC patients, which included clinical information and risk score. Enrichment analysis revealed that driver genes were involved in regulating oncogenic processes, including the cell cycle and metabolic pathways, which were associated with the progression of HCC. ssGSEA and xCell showed differences in immune infiltration and the immune microenvironment between the two risk groups. Conclusion: The 4-driver gene signature is closely associated with the survival prediction of HCC and is expected to provide new insights into targeted therapy for HCC patients.
