Comprehensive Bioinformatic Analysis Reveals an Autophagy-related Gene Signature for Predicting Outcome, Immune Status, and Drug Sensitivity in Hepatocellular Carcinoma.

Abstract

Hepatocellular carcinoma (HCC) is one of the most common malignancies in the world, but molecular complexity and tumor heterogeneity make predictive models for HCC prognosis ineffective. Many recent studies have suggested that autophagy is important in tumor progression. Using autophagy-related genes (ARGs), we attempted to create a novel signature for individual prognosis prediction in patients with HCC. Differentially expressed ARGs (DE-ARGs) in HCC and normal samples were screened using TCGA datasets. Univariate Cox and multivariate Cox regression analyses were performed to determine ARGs related to survival in HCC. An autophagy-based signature was constructed using LASSO, and its correlation with the prognosis and the immune infiltration of HCC patients was explored. In this study, we screened 32 survival-related DE-ARGs by analyzing TCGA datasets. Functional enrichment indicated that the 32 DE-ARGs may play important functional and regulatory roles in cellular autophagy, the regulation of multiple signaling pathways, as well as in the context of cancer and neurological diseases. Based on PPI Network, we identified several hub genes. LASSO Cox regression analysis selected five genes (CASP8, FOXO1, PRKCD, SPHK1, and SQSTM1) for a novel prognostic model. AUCs of 0.752, 0.686, and 0.665 in the TCGA whole set indicated that the model accurately predicted 1-, 3-, and 5-year overall survival, respectively. Cox regression analysis showed that the five-gene signature is an independent and robust predictor in patients with HCC. The high-risk group demonstrated higher levels of immune cell infiltration and exhibited a strong correlation with the immune microenvironment and tumor stem cells. In addition, we further identified PRKCD and SQSTM1 as critical regulators involved in HCC progression. The expression levels of PRKCD and SQSTM1 genes play a crucial role in chemotherapy drug sensitivity and resistance. We introduce here a novel ARG-based predictive feature for HCC patients. Effective use of this signature will aid in determining a patient's prognosis and may lead to novel approaches to clinical decision-making and therapy.