Integrative machine learning and multi-omics analysis reveals ATIC as a promoter of hepatocellular carcinoma progression.

Autophagy plays a non-negligible role in the progression and immune regulation of hepatocellular carcinoma (HCC). An integrated analysis of the autophagy-related genes (ARGs) is of significance to deepen our mechanistic understanding about HCC pathology. In our present work, based on the expression patterns of 221 ARGs, we first identified 2 autophagy-related subtypes for TCGA-HCC patients using consensus clustering method. The two subtypes showed considerable distinctions in terms of molecular characteristics, immune landscapes and clinical outcomes. To augment the clinical utility of the subtyping system, a four-gene prognostic model including ATIC, RHEB, TMEM74 and PRKCD was developed and verified through LASSO and multivariate Cox regression analyses. ROC(AUC) analysis confirmed the predictive efficacy of the model across the training and validation cohorts. Notably, HCC patients in the high-risk group exhibited elevated tumor mutation burdens and higher expression of multiple immune checkpoint genes, suggesting distinct immune-related features between risk groups. Furthermore, single-cell RNA sequencing analysis revealed that the model's marker gene-ATIC was predominantly expressed in tumor cells and proliferative T cells, with its expression showing strong and positive associations with autophagy activity. Finally, in vitro experiments were conducted to explore the potential role of ATIC in HCC. The results indicated that ATIC knockdown was associated with reduced proliferative and migratory capacities of HCC cells, along with alterations in autophagy-related phenotypes, including decreased autophagic flux. Taken together, our study provides a preliminary autophagy-related prognostic signature and identifies ATIC as a potential regulator of HCC progression.