An integrative multi-modal computational framework identified hub prognostic targets and natural inhibitors for tobacco-specific nitrosamine-induced lung adenocarcinoma.

Lung adenocarcinoma (LUAD), a major global health burden, has a poor prognosis in advanced stages. Tobacco-specific nitrosamines (TSNAs) are key carcinogens, but their systematic molecular networks driving LUAD progression remain unclear. Using a multi-modal computational approach, this study first analyzed LUAD's macro-epidemiological burden based on GBD data. Network toxicology, bioinformatics, and SHAP interpretable modeling were then applied to screen TSNA-LUAD targets, build a SHAP-interpretable Cox prognostic model, and identify hub genes. Finally, hub targets were used for molecular docking-based virtual screening of a traditional Chinese medicine library, followed by molecular dynamics simulations to analyze potential inhibitors. The GBD results indicated significant associations between lung cancer burden and gender, age, region, and sociodemographic index, confirming smoking and second-hand smoke as key risk factors. The severe burden among older populations in high-SDI regions was closely linked to the lag effect of smoking-induced carcinogenesis. A molecular interaction network of TSNAs-LUAD was constructed, identifying 91 immune-related targets enriched in key lung cancer pathways, which appear to shape an "immune-active but exhausted" tumor immune microenvironment in high-risk patients, thereby fostering immunosuppression and poor prognosis. An eight-gene prognostic model (MET, GDF15, S100P, etc.) effectively stratified patient risk, with SHAP analysis highlighting MET as the most influential driver of poor prognosis. Furthermore, eight lead compounds (resibufogenin, psoralidin, icaritin, ononin, salidroside, polydatin, hesperetin, and berberine hydrochloride) were screened as potential natural inhibitors. By elucidating the multi-target pathogenic role of TSNAs in LUAD from macro-epidemiology to molecular mechanisms, this study established an interpretable prognostic model for risk assessment and identified potential natural lead compounds for LUAD therapy.