Multi-modal investigation of doxorubicin-induced osteoporosis: Network toxicology, machine learning, and imaging techniques

Osteoporosis (OP) is a systemic skeletal disorder characterized by reduced bone mass and increased fracture risk. Emerging evidence indicates that the chemotherapeutic agent doxorubicin (DOX) significantly disrupts bone remodeling homeostasis. Our study employed an integrative computational approach to uncover the mechanisms underlying DOX-associated OP pathogenesis. Network toxicology was used to screen molecular targets in DOX-induced OP and establish protein-protein interaction (PPI) networks. The biological roles of hub genes were explored through KEGG pathway and GO analyses. Three machine learning algorithms (support vector machine-recursive feature elimination (SVM-RFE), random forests (RF), and LASSO regression) were applied to identify core genes involved in OP pathogenesis. Molecular docking studies were used to characterize the binding potential of DOX with OP-related core targets. A total of 198 overlapping candidates were identified by intersecting the 8297 candidate targets of DOX with 816 OP-associated targets. Through PPI analysis, the top 20 hub genes were determined by both maximal clique centrality (MCC) scoring and degree centrality, with 16 genes common to both centrality measures. The 16 hub genes were associated with the Jak-STAT ( P =9.1×10 −4 ) and IL-17 ( P =1.7×10 −4 ) signaling pathways. Machine learning identified six OP-associated core candidates: IL13, ZWINT, RRM2, PCNA, TYMS, and CD34. DOX exhibited high-affinity binding to IL13, RRM2, PCNA, and TYMS (ΔG < −7.0 kcal/mol). Our study suggests that DOX may induce OP through synergistic dysregulation of inflammatory and proliferative pathways, providing testable hypotheses for understanding chemotherapy-induced bone loss.