CELL INTELLIGENCE · 肿瘤细胞治疗研究

Canines share close environmental and physiological homologies with humans, making them essential large-animal models in preclinical drug discovery. However, species-specific discrepancies in therapeutic responsiveness often limit the direct extrapolation of canine data to human clinical scenarios. Although osteosarcoma (OS) spontaneously occurs in large-breed dogs at an incidence 30-fold higher than in humans, the underlying interspecies molecular differences governing tumor behavior and drug sensitivity remain poorly understood. This study aimed to systematically evaluate the translational reliability and limitations of the canine OS model by comparing human and canine malignant architectures at baseline and during anti-cancer drug exposure. By integrating single-cell RNA sequencing (scRNA-seq) from treatment-naive patient specimens with multi-conditional temporal bulk RNA-seq of representative cell lines, we interrogated their dynamic transcriptomic responses to Trabectedin. Integrated single-cell analysis revealed that human OS possesses a highly primitive, high-stemness subpopulation hierarchy and a more primed tumor-infiltrating lymphocyte (TIL) profile than canine OS. Pharmacogenomic assays demonstrated that these distinct cellular architectures dictate therapeutic vulnerabilities: The rigid, high-stemness hierarchy of human OS facilitates transcriptomic robustness and adaptive drug resistance. In contrast, the fluid, less primitive network of canine OS renders its tumor cells prone to immediate, stochastically driven functional network collapse and superior Trabectedin sensitivity. Collectively, our study provides a comprehensive comparative molecular framework for human and canine OS. These findings emphasize the necessity of accounting for species-specific stemness and immune landscapes, ultimately accelerating the precision development of next-generation OS therapeutics.