Endocrine, nutritional and metabolic diseases drive cancer risk through subtype-specific and shared proteomic mediators

Cancer is intricately linked to metabolic disorders, yet a comprehensive atlas delineating the impact of endocrine, nutritional and metabolic diseases (ENMs) on cancer trajectories remains unexplored. Here, leveraging an innovative multi-state model, we systematically mapped the influence of 25 ENMs on cancer trajectories—from primary cancer incidence to multi-primary cancer and mortality—within a large-scale prospective cohort of 405,023 participants. Our findings reveal heterogeneous effects of ENMs across cancer sites and types, identifying 8 cancer-related ENMs classified into four distinct subtypes: C1 (iron disorder-dominated), C2 (type 2 diabetes-dominated), C3 (lipid/lipoprotein disorder-dominated), and C4 (osteoporosis-dominated). Notably, subtype-specific and subtype-shared proteomic mediators were delineated through rigorous multi-step selection. Parallel mediation analyses quantified the contribution of proteins to ENM-driven oncogenesis, with RNASET2 emerging as the predominant mediator across all subtypes. Furthermore, a deep-learning model integrating ENM subtypes and proteomic mediators enables dynamic, personalized prediction of cancer incidence (AUC=0.80–0.90), outperforming conventional Cox models. Our findings establish ENMs as modifiable cancer determinants, provide mechanistic insights into metabolic dysregulation-driven tumorigenesis, and deliver a translatable tool for personalized cancer prevention.