Machine learning-assisted multi-dimensional transcriptomic analysis of cytoskeleton-related molecules and their...

Hepatocellular carcinoma (HCC) is a leading cause of cancer-related death worldwide, with a poor prognosis due to its aggressive nature and limited treatment options. Cytoskeletal dynamics play a critical role in tumor progression, but the prognostic and therapeutic potential of cytoskeleton-related genes in HCC remains underexplored. In this study, transcriptomic data from the TCGA-LIHC dataset were used to identify differentially expressed cytoskeleton-related genes associated with overall survival (OS). Prognostic models were constructed using LASSO regression and random forest algorithms, and validated in two independent cohorts (ICGC LIRI-JP and CHCC-HBV). Single-cell sequencing (scRNA-seq) and spatial transcriptomics analyses explored the expression and functional roles of key genes, while drug screening and molecular docking identified potential therapeutic agents, followed by in vitro and in vivo validation. The analysis identified 110 cytoskeleton-related DEGs, with 13 significantly associated with OS. A robust five-gene prognostic model (ARPC1A, CCNB2, CKAP5, DCTN2, TTK) was developed using LASSO regression and validated across cohorts. The model was integrated into a clinical nomogram, demonstrating good calibration and utility. Single-cell and spatial transcriptomics revealed high expression of the five genes in malignant tissues and their association with immunosuppressive microenvironments. High-risk scores correlated with TP53 mutations. Drug screening identified irinotecan and sorafenib as potential agents targeting TTK, with combined treatment significantly inhibiting tumor growth in vitro and in vivo. This study highlights the prognostic and therapeutic significance of cytoskeleton-related genes in HCC. The five-gene model provides a reliable tool for risk stratification, and the irinotecan-sorafenib combination shows promise as a therapeutic strategy.