Integrated bibliometric and machine learning analyses identify candidate autophagy related therapeutic targets for prostate cancer.

PURPOSE: Autophagy is increasingly recognized as a promising therapeutic target in prostate cancer (PCa), yet its research landscape has not been systematically evaluated. This study uses bibliometric methods to characterize global research trends and identify emerging hotspots in autophagy-related PCa studies. METHOD: Publications from 2003 to 2025 were retrieved from the Web of Science Core Collection. VOSviewer, CiteSpace, and ScimagoGraphica were used to construct collaborative networks, research frontiers, and development trends. Additionally, clinical trial literature was sourced from PubMed to assess the clinical progress of autophagy-targeted therapy for PCa. The GSE46602 dataset was also employed alongside machine learning techniques to identify autophagy-related genes in PCa. RESULTS: A total of 1,863 publications were identified, contributed by 11,161 authors across 79 countries. The United States and China were leading contributors, with Paul B. Fisher as the most productive author and the National Natural Science Foundation of China as the top funding agency. Current hotspots include epithelial-mesenchymal transition, ferroptosis, drug resistance, non-coding RNA, and berberine. Machine learning analysis identified five core autophagy-related genes: CCL2, CTSB, DAPK1, MYC, and VEGFA. ROC analysis in both training and validation cohorts demonstrated that CTSB, DAPK1, and VEGFA exhibited AUC values greater than 0.75, indicating strong diagnostic potential. CONCLUSIONS: Global research on autophagy in PCa is rapidly expanding, especially in ferroptosis crosstalk, RNA regulation, natural compounds, and resistance mechanisms. These findings highlight autophagy as a promising biomarker and therapeutic target in PCa.