IDDF2024-ABS-0182 Exploring the role of mitochondrial dynamics-related genes in the molecular classification and prognostic value of liver hepatocellular carcinoma

Background

Liver hepatocellular carcinoma (LIHC) is characterized by elevated incidence and mortality rates. Mitochondrial dynamics represents a critical facet of functions in apoptosis resistance and invasion ability of cancer cells. However, the role of mitochondrial dynamics in LIHC remains unclear. The study aims to explore the role of mitochondrial dynamics-related genes (MDRGs) in LIHC, identify molecular subtypes, and develop a novel prognostic signature for LIHC.

Methods

The MDRGs were extracted from the MitoCarta3.0 database and the gene expression data were from TCGA and GEO databases. We used the R package 'limma' to determine differentially expressed genes. The prognostic value was evaluated using univariate Cox analysis. Molecular clusters were recognized by the R package 'ConsensusClusterPlus'. The immune infiltration was analyzed using the TIMER algorithm. The key MDRGs were screened using a random forest (RF) algorithm and the hub MDRG was selected via the cytoHubba algorithm. The TCGA cohort was split 7:3 into train and internal validation sets; an independent cohort (GSE54236) was the external validation set. Forty-two prognostic models were generated by integrating multiple algorithms, including random survival forest (RSF), Coxboost, etc.

Results

A total of 10 key MDRGs were collected. Based on these genes, we identified two LIHC clusters (C1 and C2), with C1 showing higher MDRGs expression and poorer prognosis than C2. Moreover, significant differences in immune infiltration were detected between the two clusters. The RF algorithm ultimately selected 6 MDRGs for the construction of models. We finally constructed 42 prognostic models and identified a model using the RSF algorithm that exhibited the best predictive ability (mean AUCs for the internal validation set and external validation set were 0.727 and 0.720, respectively). Finally, CHCHD3 was selected by cytoHubba as a hub gene. Knocking down CHCHD3 markedly inhibited the migration of tumor cells in Huh7 and HepG2 cell lines.

Conclusions

The MDRGs played a crucial role in LIHC. We identified two distinct mitochondrial dynamics-related clusters of LIHC; high MDRGs expression of cluster had poor prognosis. The prognostic model developed based on MDRGs showed satisfactory performance. Our findings suggested the promising effect of MDRGs in the management and treatment of LIHC.