CELL INTELLIGENCE · 肿瘤细胞治疗研究

Adoptive T-cell therapies have revolutionized hematological oncology, yet their robust efficacy frequently encounters significant translational bottlenecks in osteosarcoma (OS). The marked discrepancy between potent in vitro cytotoxicity and clinical failure highlights a critical need to re-evaluate therapeutic assessment. Conventional preclinical platforms, disproportionately reliant on 2D cultures and subcutaneous immunodeficient xenografts, strip away the formidable anatomical, metabolic, and immunological barriers intrinsic to the OS microenvironment, introducing significant translational bias. By selecting studies that detail mechanistic interactions within the tumor-host ecosystem, this narrative critical review appraises OS T-cell therapies strictly through the lens of preclinical model evolution. Moving away from a linear hierarchy of complexity, we critically evaluate these platforms as a complementary, fit-for-purpose framework-from advanced 3D spatial matrices assessing spatial infiltration to high-fidelity orthotopic and spontaneous pulmonary metastasis models evaluating biodistribution and metastatic control. We further interrogate the utility of patient-derived xenografts in preserving authentic heterogeneity, fully immunocompetent syngeneic platforms for decoding systemic tumor-host immune networks, and spontaneous large-animal canine models integrated with multimodal in vivo imaging to bridge the macroscopic anatomical gap. Ultimately, conquering OS requires moving beyond the effector cell to modulate the entire tumor-host ecosystem. By establishing how high-fidelity models evaluate preclinical antitumor activity and synergistic mechanisms of combinatorial interventions, this review underscores the critical transition from oversimplified assays to physiologically relevant platforms, guiding precision immunotherapies toward more rigorous preclinical evaluation and improved translational potential.