Autophagy Associated Genes (ARGs) -Based Predictive Model AIDPS for Prostate Cancer

Prostate cancer (PCa) is one of the most prevalent cancers in men globally. Autophagy-related genes (ARGs) are believed to play crucial roles in various biological processes associated with PCa. This study aimed to identify and assess autophagy-related features for predicting clinical outcomes in PCa patients. Data from single-cell sequencing and RNA sequencing were sourced from public GEO and TCGA databases. Cells were clustered and annotated through dimension reduction analysis. Epithelial cells, T cells, and fibroblasts were isolated to explore their heterogeneity. ARGs were obtained from the HADb database. Survival analysis was performed using Kaplan-Meier (K-M) curves, and a prognostic risk model was developed using 101 machine learning algorithms. Additionally, gene colocalization and Mendelian randomization analyses were conducted. Univariate Cox analysis was used to identify prognostic genes from differentially expressed genes (DEGs) and ARGs in each dataset. The risk model was generated using an artificial intelligence-derived prognostic signature (AIDPS), which outperformed existing models in predicting PCa prognosis across all datasets. Patients in the high-risk group had significantly poorer disease-free survival (DFS) compared to those in the low-risk group (all p < 0.05). The best performing model was Ridge regression (C-index = 0.726). We also observed significant differences in IC50 values for Dactinomycin_1811, Dactolisib_1057, Luminespib_1559, and Paclitaxel_1080 between the risk groups. A significant association was found between prostate hyperplasia and prostate cancer at SNP sites rs2743987 and rs7768988. Our ARG-based predictive model, AIDPS, is a reliable and effective tool for predicting prognosis and guiding treatment in prostate cancer.