Abstract
INTRODUCTION: With the increasing impact of hepatocellular carcinoma (HCC) on society, there is an urgent need to propose new HCC diagnostic biomarkers and identification models. Histone lysine lactylation (Kla) affects the prognosis of cancer patients and is an emerging target in cancer treatment. However, the potential of Kla-related genes in HCC is poorly understood. METHODS: A variety of machine learning methods were used to construct and validate a model of differentially expressed Kla genes with comprehensive evaluations included ROC, Kaplan‒Meier curve, Cox regression, decision curve. Immune infiltration gathered with spatial transcriptome was performed using integrated data from multiple databases. Furthermore, single-cell analysis was used to discover the cell-cell communication and Mendelian randomization was used to study the causal relationships between immune cell and HCC. Lastly, qRT-PCR was used to verify the expression of Kla genes. RESULTS: We established a model consisting of 12 genes that had well prognostic performance and were identified as independent prognostic factors. Single-cell analysis showed that CD8 T+ cells and conventional dendritic cells were enriched in HCC patients. Spatial transcriptomics analysis indicated that the Kla genes influenced the immune characteristics of HCC. Mendelian randomization results showed that TBNK and monocytes were the main risk factors. qRT-PCR validation results indicated that the expression of multiple genes in Huh7 cells was significantly higher than in LO2 cells. CONCLUSION: Overall, a Kla-related model was established, which may provide new strategies and insights for the treatment and diagnosis of HCC.