Deciphering diverse cell‐death patterns to predict the prognosis and potential therapy target of hepatocellular carcinoma patients

Abstract Ninety percent of all primary liver malignancies are hepatocellular carcinomas (HCC), making liver cancer the third most common cause of cancer‐associated mortality. Different patterns of programmed cell death (PCD) are crucial for the survival of tumors, and they might serve as a prognostic marker for HCC. The construction of the prognostic models involved an analysis of eighteen PCD patterns: apoptosis, cuproptosis, entotic cell death, netotic cell death, parthanatos, pyroptosis, lysosome‐dependent cell death, autophagy‐dependent cell death, alkaliptosis, oxeiptosis, NETosis, immunogenic cell death, ferroptosis, ANOIKIS, disulfidptosis, necroptosis, macroautophagy, and methuosis. Bulk and single‐cell transcriptome, genomics, and clinical parameters were analyzed in training. Additionally, three datasets were used as validation cohorts. The hub gene in the model, SQSTM1, was discovered, and in vitro and in silico studies were performed to learn how SQSTM1 operated in PCD. A 16‐gene cell death index (CDI) was developed using machine learning. Validation across four datasets shows CDI scores distinguish cancer from adjacent tissues, with higher scores correlating to worse prognosis. Further spatial and single‐cell transcriptome investigations reveal that SQSTM1 is linked to multiple cell death pathways and interacts with inhibitory M2 macrophages. Silencing SQSTM1 induces apoptosis and increases autophagy and ferroptosis gene expression, limiting tumor growth and identifying SQSTM1 as a potent biomarker for HCC diagnosis. In summary, we created a unique CDI signature by thoroughly involving the greatest number of cell death patterns to date, eighteen, which might be effective to predict clinical prognosis and provide a novel treatment option.