Abstract

Background Lung adenocarcinoma (LUAD) evades regulated cell death, rendering PANoptosis a promising target. We investigated whether the antitumor flavonoid astragalin suppresses LUAD via PANoptosis and identified associated prognostic signatures. Methods The transcriptome data for LUAD patients were obtained from The Cancer Genome Atlas (TCGA) database, with PANoptosis-related genes (PARGs) and astragalin targets predicted via network pharmacology. Differentially expressed genes (DEGs) were used as input for weighted gene co-expression network analysis (WGCNA) to identify co-expression modules. Next, we integrated the overlapping genes to identify and prioritize a prognostic gene signature via ensemble machine learning [least absolute shrinkage and selection operator (LASSO), Boruta, extreme gradient boosting (XGBoost)]. Functional enrichment, immune infiltration, and drug sensitivity prediction were performed across different risk groups. Key targets were assessed for potential interaction with astragalin through molecular docking simulations, quantitative real-time polymerase chain reaction (qRT-PCR), western blotting, and in vitro assays. Results Intersection with WGCNA and PARGs and astragalin-related genes yielded 59 candidates enriched in cell cycle, p53 signaling, and immune pathways. Integrating protein-protein interaction (PPI) and machine learning, we prioritized a four-gene signature ( TOP2A , PLK1 , CDKN3 , and CCNB1 ) consistently upregulated in LUAD and predictive of poor prognosis. High-risk tumors showed suppressed p53/autophagy but elevated M1 macrophage infiltration, with all four genes positively correlated with M1 abundance. Molecular docking predicted strong astragalin binding to these targets (−7.4 to −8.5 kcal/mol), and in A549 cells, astragalin treatment led to the downregulation of these genes and induced mitochondrial apoptosis via Bcl-2 suppression and Bax/caspase-3 activation. Conclusions Our study identifies a four-gene cell cy