建立转化研究生态系统：一个全球联网的PDX模型平台以加速肿瘤药物开发

Background: Patient-derived xenograft (PDX) models continue to play a critical role in translating novel oncology drug innovation from discovery through late-stage development. The START Center for Cancer Research is a global translational and clinical trials oncology network that includes XenoSTART, a PDX development and testing division. XenoSTART offers end-to-end capabilities, beginning with the establishment of PDX models directly from conventional and trial patients across diverse indications, reflecting current treatment landscapes with the added ability to support a full suite of in vivo study capabilities, provide models and data under license, and tailored model development. Strategic partnerships allow for extended service capabilities which include orthotopic imaging, radioligand therapy (RLT), and humanized systems capabilities, enabling mechanistic depth and modality-specific translational insights rarely available through a single platform. Methods: XenoSTART PDX (XPDX) models are collected from primary or metastatic patient tumor samples and engrafted into immunocompromised mice under standardized workflows; a curated extraction of clinical details and treatment histories from donor patients ranging from newly diagnosed through heavily pretreated is performed for all collected samples. Resulting models are serially passaged and further developed until growth stabilization. Established models are profiled using integrated molecular and pathological analysis including WES and RNAseq, receptor expression, and advanced bioinformatics to support biomarker discovery or mechanistic insights and further characterized through in vivo responses to standard-of-care and emerging therapies. Bioinformatic analyses are conducted using validated pipelines to evaluate molecular signatures and biomarker associations. In vivo studies following harmonized protocols align with clinically relevant dosing schedules. Results: The XenoSTART platform generates a diverse and deeply characterized XPDX repository reflecting contemporary treatment landscapes, with high rates of molecular and phenotypic fidelity. Treatment benchmarking replicates known clinical response patterns to standard-of-care agents. Integrated clinical and molecular datasets revealed biomarkers associated with treatment sensitivity and resistance. Specialized collaborations enable innovative translational studies including orthotopic imaging, RLT, and humanized immune-oncology evaluation. Conclusion: XenoSTART's globally connected PDX ecosystem provides a clinically focused, best-in-class translational resource that enhances predictive accuracy, informs patient-stratification strategies, and drives more confident decision-making from early discovery through late-stage clinical development.